How much

productivity
can GenAI
unlock in India?
The AIdea of India: 2025

The AIdea of India: 2025 1

 Foreword
Contents
Chapter 1
Generative AI:
Shaping tomorrow
Executive
summary
Chapter 2
Pivoting to
AI-first digital
transformation

Page no:32
Page no: 14
Page no: 06
Page no: 04

2 The AIdea of India: 2025

 Chapter 3
Chapter 6
Transforming work
with GenAI Policy agenda
for India

Chapter 4
Industries in
transformation Annexures

Chapter 5
Page no: 48

Government
Services
Page no: 58

Page no: 124

Page no: 116
Page no: 110

The AIdea of India: 2025 3

Foreword
O
ver the past few years, innovation possible to use open-source models for as
in Generative AI (GenAI) has low as a few thousand rupees a month
progressed at an extraordinary in India.
pace, reaffirming its transformative Yet, amidst all this innovation, enterprise
potential across a number of domains. adoption rates of GenAI remain very low.
The possibilities are vast and hold the Our survey shows that 36% of Indian
promise of profound changes on the enterprises have allocated budgets and
horizon for millions of Indian citizens. begun investing in GenAI, while another
Technical breakthroughs have been jaw 24% are testing its potential. Technology
dropping. We have quickly moved from sector clients are leading the way, with
auto-complete chatbots to reasoning Life Sciences and Financial Services
machines capable of spinning out following suit. Despite this, the business
credible, human like ‘Chains of Thought’ value remains limited, with just
(CoT) to find solutions to complex 15% having GenAI workloads in
problems. Today, multi-modal large production and only 8% able to fully
language models (LLMs) can enable measure and allocate AI costs.
seamless processing of text, audio, This is not surprising – it takes time for
image, and video. Emerging trends like innovation to be packaged and made
Agentic AI are enabling autonomous ready for enterprise adoption. Enterprises
entities capable of taking actions. The need clarity on ROI and guarantees
evolution of new hardware platforms around issues like hallucination, data
and new AI accelerators has ensured privacy and algorithmic bias as they craft
the computational power to support their digital transformation roadmaps.
increasingly sophisticated models, Over the next few years, we expect an
having even a trillion parameters and explosion of enterprise adoption as these
groundbreaking efficiency. issues are addressed and AI and GenAI
Along the way, the cost of intelligence models make their way into the
has fallen, driven by the open-source enterprise mainstream.
movement and the trend to use purpose Just as during the earlier era of mobile
specific small language models (SLMs). disruption, fintech and healthtech
This is making AI accessible to smaller enterprise adoption will lead to the birth
businesses and very soon it may become

4 The AIdea of India: 2025

of AI-first companies with new business potential as the use case and data capital
models and revamped economics. These of the world. The focus will need to be on
firms will compete with digital interfaces enhancing data accessibility and compute
powered by chat, voice and regional infrastructure, fostering AI research
language models. Algorithms and new and innovation through initiatives like
datasets will help drive population-scale localized LLMs, and addressing challenges
operations. AI-driven apps will transform in responsible governance, intellectual
knowledge work. property rights, and data protection.
On the other hand, we need to address
The coming wave of change has
the coming potential job dislocation in
significant implications for India.
the workforce by implementing
In industries like financial services,
aggressive skilling programs and
healthcare and retail, we expect AI
apprentice schemes.
to reshape basic processes including
customer acquisition, operations and This report is an in-depth exploration
service. Industries including IT/ITeS and of GenAI’s current state in Indian
BPO will undergo more dramatic changes. enterprises, key trends shaping its
Next-generation industries like biotech, future, and implications for Indian
advanced manufacturing and renewables enterprises and policymakers.
will have the potential to leapfrog to I hope you find this report valuable
AI-first business models. - happy reading!
Our analysis reveals that, at a macro
level, the AI platform shift will impact
38 million employees, potentially
driving a 2.61% boost in productivity
by 2030 in the organized sector.
Enterprises will need to reorient Rajiv Memani
themselves rapidly to deal with this
coming impending tides of change. Chairman and CEO,
EY India
There will also be significant pressure
on India’s policy agenda. On one hand,
there is the imperative to realize India’s

The AIdea of India: 2025 5

6 The AIdea of India: 2025
Executive
summary
T
he entire earth will be be paired with practical applications that solve
converted into a huge real-world problems, empower users, and bridge gaps in
brain, as it were, capable digital access and infrastructure.
of response in every one of its
parts.” This was Nikola Tesla, in
1904, predicting the impact of Innovation in GenAI continues at
the radio on the world.
a scorching pace
Every generation believes
it stands on the brink of
transformation, fueled by the Innovation in GenAI surged in 2024, marking a
technologies of its time. Today, transformative year for the technology.
as we contemplate the AI era, There was rapid progress in Multimodal AI, integrating
it feels like one of those pivotal text, images, audio and video into unified models that
moments. On one hand, there significantly enhance real-world usability. This was
is exponential innovation — particularly evident in the incorporation of these models
AI’s promise is vast, with the into AI-powered phones and emerging form factors like
potential to revolutionize smart glasses, enabling seamless and intuitive interactions
industries, redefine work, across diverse applications.
and unlock unprecedented
creativity and productivity. The open-source movement gathered steam. Leading
Breakthroughs in GenAI have open-source large language models (LLMs) such as
been astounding, and the Meta’s Llama 3 and Mistral Large set new benchmarks for
possibilities appear limitless. performance while addressing critical concerns about data
privacy and security. Simultaneously, there was a growing
Yet, there is the critical challenge realization that smaller, domain-specific models could
of making this transformation often outperform their larger counterparts in
relevant and accessible to targeted tasks.
consumers and enterprises. For
AI to truly deliver on its promise, Year 2024 also saw breakthroughs in reasoning. Models
cutting-edge innovation needs to such as OpenAI’s GPT-4o31, and Google’s AlphaProof2

1. https://openai.com/index/deliberative-alignment/
2. https://www.ebi.ac.uk/training/online/courses/alphafold/inputs-and-outputs/a-high-level-overview

The AIdea of India: 2025 7

GenAI in India: The current state of play
EY India’s C-suite GenAI survey
We conducted an in-depth GenAI survey covering more than 125 C-suite executives across India. They
represent diverse sectors, including Financial Services, Retail, Life sciences, Media and Entertainment,
Technology, Automotive, Industrials and Energy.

GenAI journey GenAI strategy: Direction and alignment

Integration with existing software means enterprises’ More than half of the enterprises have a GenAI strategy
exposure to GenAI is high. However, only a few have the but only some have a fully integrated strategy with clear
technology in production. execution plans

Fully integrated strategy

12% 22% with clear execution plans
36% Strategy aligned with
11% business goals, but
34% 15% 18% execution plans are lacking

8% Strategy exists, but not

30% 9% aligned with business
goals
Basic understanding, no
POCs completed Productionalization in progress formal strategy
POCs in progress AI adopted
No POCs done No clear impact 39% No clear strategy

Architecture: GenAI platform and Implementation: Buy versus build

integration approach approach
Architecture integration is limited and enterprises are Approximately one in four have defined approach but
looking at ways to increase application application is uneven
11% Fully integrated and 16% 10% Well-defined and
optimized platform consistently applied
19% approach
Integrated architecture in
4% place; facing utilization Defined approach, but
challenges not consistently applied
Platform selected; Preliminary approach, not
integration just started fully defined or executed
Platforms identified, but 21% Aware of options, but no
no integrated architecture clear decision framework
43% No platform selected /
32% No consideration of buy
23% integration approach 21% vs. build yet
defined

Data: Platform readiness for GenAI Talent: Resource availability for GenAI
adoption adoption
Enterprises in India are at different stages of data AI expertise is a key need for most enterprises as they
readiness, with only a few at a mature level undergo GenAI transformation
3% 16% Fully ready and mature
3% 16% Extensive expertise and
Mostly ready, minor gaps resources for effective
deployment
Partially ready, requires Talent exists but
enhancements insufficient to support all
initiatives
Needs significant
23% improvements Have some skills but need
significant investment
Not ready 22% Aware of skills but lack
them and have no
acquisition strategy
19% 39% Have not thought about
24% 35% specific GenAI skill
requirement

8 The AIdea of India: 2025

achieved remarkable progress in solving GenAI in India: Shaping tomorrow
complex problems across disciplines like science,
mathematics and programming, consistently
surpassing previous benchmarks. These advanced India will chart a unique path as this technology
capabilities started to get packaged into agentic AI evolves. We see five key trends that will
systems which aim to independently plan, reason, significantly influence India’s AI evolution.
and execute tasks by dynamically leveraging tools
and resources. Though still in its infancy, this
agent-driven paradigm promises to fundamentally
reshape our understanding of work and the way we 01 Chat, voice, regional languages
augment digital interfaces
design software systems.
Hardware innovations continued to underpin these
02
Agents enable the transformation
advancements in GenAI. NVIDIA maintained its
of knowledge work
leadership with the Blackwell platform, enabling
trillion-parameter models while competitors drove
significant breakthroughs in AI accelerators.

03 LLMs are not all you need:

Toward compound AI systems
Moving from demos and labs to
enterprise grade capabilities
04 The falling cost of AI
Yet, despite these breakthroughs there is also
increasing doubt about the pace and magnitude
The evolution of an
of the impact of GenAI. Goldman Sachs, for
instance, has highlighted the imbalance between
the massive investments being funneled into AI and
05 Indic AI ecosystem

the uncertain returns. In a June 2024 report titled

“Gen AI: Too Much Spend, Too Little Benefit?”,
the firm projected that tech giants and other
‘good enough’ for scaling across many use cases.
companies are set to invest nearly US$1 trillion
Our survey of Indian enterprises suggests that
in AI-related expenditures over the coming years,
customer service, operations and sales & marketing
spanning data centers, specialized hardware, and
functions are already leading the way in adoption.
infrastructure upgrades. Despite these staggering
Over the next few years, as these teething issues
sums, the tangible benefits remain elusive.
are addressed, AI and GenAI models make their
Our survey of Indian enterprises suggests that 36% way into the enterprise mainstream across all
of enterprises have budgeted and started investing functions and departments.
in GenAI while another 24% are experimenting with
it. Technology sector clients have been leading
the way with Life Sciences and Financial Services AI augmented interfaces will transform
following suit. At the same time business value
consumer apps
delivered is relatively low with only 15% of Indian
enterprises report having GenAI workloads in AI-powered chat, voice and regional language
production, and just 8% being able to fully measure tools are already making an impact and this trend
and allocate AI costs. will accelerate as digital models diffuse across
the Indian consumer, enterprise and government
This is not surprising. Packaging innovation into
landscape. GenAI native interfaces will also serve
products and services that enterprises can use
as front doors to onboard less digitally savvy users
is a time-consuming process. Enterprises need
into the digital economy. Solutions like NPCI’s Hello!
clarity on ROI and guarantees around issues like
UPI and IRCTC’s AskDisha chatbot demonstrate
hallucination, data privacy and algorithmic bias as
this shift, enhancing inclusivity for underserved
they craft their digital transformation roadmaps.
populations in semi-urban and rural areas.
Rapid advancements to date have already made AI

The AIdea of India: 2025 9

Agents will transform knowledge work A rich Indic AI ecosystem will evolve to
The rapid integration of AI Agents into sectors cater to unique Indian needs
like information technology, finance, customer There has already been a mushrooming of Indic
service and healthcare will reshape traditional LLMs that leverage open-source models fine-tuned
ways of working, presenting both opportunities with Indian language datasets. A key initiative in
and challenges for Indian professionals. Our this space is Bhashini, a government-led AI project
analysis (more on this in ‘Transforming work with aimed at creating an open-source Indic language
GenAI’) indicates potentially large productivity dataset to expand internet and digital service
improvements that will begin to manifest accessibility in Indian languages. Going forward,
themselves and companies will begin to gear up to AI will increasingly become part of the India stack
help employees manage the coming transition to and available as digital public infrastructure to build
new ways of working. next generation platforms.
A burgeoning GenAI start-up ecosystem and local
Enterprises will start to move to an AI infrastructure will help drive adoption in
AI-embedded tech stack Indian enterprises.
Enterprises will learn to treat LLMs as but one part
of an evolving AI enabled tech stack. AI adoption
will accelerate as enterprises integrate LLM Pivoting to AI-first digital
capabilities with classical AI techniques, new transformation
modes of automation and the emerging modern
data stack.
Similar to the transformative impact of the digital
revolution, the accelerating shift toward AI-
AI costs will continue to fall driven platforms is poised to reshape every factor
The cost of using AI models has already influencing a company’s EBITDA. Across Indian
plummeted, making them increasingly accessible to enterprises, AI-first approaches are steadily taking
enterprises. OpenAI’s GPT API costs, for example, root, embedding themselves throughout the value
have dropped nearly 80% in two years, while chain to enhance operational efficiency and unlock
open-source releases like Meta’s Llama are new avenues of value creation.
unlocking new capabilities. This cost is expected At a foundational level, AI automates workflows,
to fall to around INR120 per hour* or lower as detects patterns, and delivers real-time predictions,
India specific LLMs offerings become viable. creating a closed-loop system for continuous
(*Assuming that the cost is US$4 per million tokens learning. This will help companies optimize value
and the application uses 100 tokens per second chains, enhance revenue streams through improved
continuously, the enterprise would spend channels and pricing, and transform delivery with
US$1.44 per hour.) new interfaces.

10 The AIdea of India: 2025

The agenda for enterprises

Reimagine the Rethink the tech Move to AI-ready Getting your Confronting the
business and stack data people ready for AI changing frontier of
operating model risk

A new AI-powered tech stack is emerging, and workforce adaptability. Change management
combining foundational models with specialized bridges the gap between innovation and
tools. Enterprises are increasingly adopting execution, enabling organizations to thrive
SLMs for domain-specific tasks due to their cost in an AI-driven world.
efficiency, precision, and ability to run on edge
AI-first strategies introduce risks related to bias,
devices. Enterprise software providers such as
cybersecurity and explainability. Organizations
SAP, Salesforce and Oracle are embedding AI
are mitigating these by adopting automated
into their platforms, accelerating adoption with
compliance systems, real-time anomaly detection,
ready-to-deploy AI tools. Meanwhile, traditional
and explainable AI models. Regulatory frameworks
Robotic Process Automation (RPA) is evolving
like India’s Digital Personal Data Protection
into intelligent automation by integrating GenAI,
Act (DPDP Act 2023) further emphasize the
enabling systems to adapt dynamically to changes
importance of responsible AI practices, especially in
without manual intervention.
sensitive sectors like healthcare and finance.
A solid data foundation is pivotal to enterprise
AI success. Enterprises are implementing robust
governance frameworks, addressing challenges Transforming work with GenAI
related to data quality, diversity and sensitivity.
Modern data stacks, including cloud platforms
and scalable data lakes, enable real-time ingestion In India GenAI has the potential to drive
and processing, essential for AI implementation. productivity gains, impacting millions of workers
Companies that nurture proprietary datasets and redefining the future of work.
are gaining competitive advantages by achieving EY conducted a study of over 10,000 tasks in
superior model performance. critical industries that contribute to the Indian
Preparing people for AI is crucial to unlocking its economy. To assess GenAI’s impact on productivity,
full potential, ensuring both technological adoption tasks were analyzed based on exposure (potential

The AIdea of India: 2025 11

impact of GenAI), complementarity (human due to its higher labor share in gross output, while
oversight needed) and intensity (frequency of tasks manufacturing and construction will see smaller
analyzed in granular time units). A ‘Productivity impacts. However, even in these sectors, AI can
Uplift’ Indicator was created, to quantify drive efficiencies through better capital deployment
this potential impact in terms of Automation and resource utilization, ultimately lowering labour
(elimination of the task), Augmentation expenses and improving overall cost efficiency.
(doing the same task better using GenAI) and
Realizing this potential requires reimagining
Amplification (enhancing the nature of the task
processes, redefining workflows and reskilling
and making it richer).
the workforce. The successful adoption of GenAI
This allowed us to analyze productivity gains at requires clear strategies, piloting use cases, and
job role, functional and organizational levels. scaling solutions, alongside reimagining processes,
Our analysis reveals that 24% of tasks can be redefining KPIs, and targeted reskilling. Large-scale
fully automated, while time spent on another upskilling initiatives, supported by public-private
42% can be significantly reduced, freeing up partnerships and AI-focused training programs, are
8-10 hours per week for corporate workers. This crucial to bridging the skill gap. With investments
translates to a productivity boost of 2.61% by in skills, data and infrastructure, GenAI can drive
2030 in the organized sector affecting 38 million economic productivity and ensure a future-ready
Indian employees and an additional 2.82% in the workforce for India.
unorganized sector. The largest productivity gains
from GenAI are expected in the services sector

Productivity gains across key sectors

This graph illustrates the labor cost as a percentage of gross output on the x-axis and the percentage
productivity improvement through AI on the y-axis. The size of the bubble represents the potential labor
efficiencies created by AI for the industry.

50%
productivity significantly

45%

IT | 19%
40%
AI enhances

Retail | 5%
Productivity potential

Banking | 9%
35%
Pharma | 2%
Insurance | 8%
30%
Telecom | 5%
Automobile | 2%
productivity marginally

25% Metals and Mining | 4%

Healthcare | 13%
20%
Media and Entertainment | 5%
AI enhances

15%

10%
0% 10% 20% 30% 40% 50% 60%

Labor cost by gross output

Labor plays Labor plays a
a smaller role larger role

EY India jobs study: Transforming work with GenAI

12 The AIdea of India: 2025

A policy agenda for India To ensure Responsible AI, the government has
prioritized transparency, fairness and safety
through consultations and oversight. Plans include
India’s AI policy landscape reflects a balanced forming a National Committee on Responsible
approach to fostering innovation while ensuring and Trustworthy AI, addressing bias, privacy and
responsible deployment. The IndiaAI Mission stands accountability. The DPDP Act requires
at the forefront, with a financial commitment businesses to adopt privacy-preserving AI tools,
of over INR10,000 crore to develop India’s AI anonymization protocols, and compliant
ecosystem across seven pillars, including access workflows, aligning AI development with evolving
to high-quality datasets, expanded compute data protection standards.
infrastructure, and responsible AI governance. Key
initiatives include establishing the India Dataset India’s strategic AI policies, anchored in inclusivity,
Platform for organized, sector-specific data access, data sovereignty and accountability, aim to position
deploying 10,000 GPUs to scale AI research, the country as a global AI leader while mitigating
and promoting AI solutions in critical sectors like risks, promoting innovation, and ensuring ethical AI
healthcare and agriculture through R&D incentives adoption across public and private sectors.
and innovation challenges.

The AIdea of India: 2025 13

Chapter 1

Generative AI:
Shaping tomorrow

14 The AIdea of India: 2025

The AIdea of India: 2025 15
Chapter 1

Generative AI:
Shaping tomorrow

The promise still holds

Multimodal AI advancements,
agent-driven systems, and hardware Over the past few years, innovation in GenAI has
progressed at an extraordinary pace, reaffirming
advancements like NVIDIA’s Blackwell
its transformative potential across a number of
are reshaping global applications, domains. The possibilities are vast and hold the
moving GenAI from labs to promise of profound changes on the horizon. In
the domain of healthcare, AI could accelerate
enterprise-grade solutions
breakthroughs in biology, enabling the rapid
development of cures for diseases like cancer and
The rise of open-source LLMs and the Alzheimer’s while extending human lifespans. In
success of smaller, domain-specific neuroscience, it offers hope for understanding
and treating mental illnesses such as depression
models are addressing privacy, and schizophrenia, while also enhancing human
efficiency, and targeted use-case needs cognition and emotional well-being. Economically,
AI promises to potentially uplift billions out of
poverty by optimizing resource distribution and
The rapidly falling costs of AI revolutionizing industries like agriculture and
solutions, like 80% drop in the price clean energy. In governance, AI might strengthen
of OpenAI’s APIs over two years, governance by enhancing public services and
reducing corruption. Finally, in education and work,
are making advanced capabilities AI can democratize knowledge access and redefine
increasingly accessible to enterprises meaningful human contributions, ensuring an
inclusive future where technology enriches, rather
than replaces, human purpose.
India is leveraging GenAI for
regional language accessibility,
digital inclusivity, and transformative But GenAI is not without its
consumer apps skeptics

As Indian enterprises adopt Yet, as with all transformative technologies, GenAI

has its share of doubters. While its promise is vast,
AI-embedded tech stacks; concerns about the pace and magnitude of its
start-ups and SaaS companies will impact linger. Goldman Sachs, for instance, has
lead the charge, driving innovation highlighted the imbalance between the massive
investments being funneled into AI and the
and integration across industries uncertain returns. In a June 2024 report titled
in the coming years “Gen AI: Too Much Spend, Too Little Benefit?”,
the firm projected that tech giants and other
companies are set to invest nearly US$1 trillion

16 The AIdea of India: 2025

 Emerging
in AI-related expenditures over the coming years,
spanning data centers, specialized hardware, and
infrastructure upgrades. Despite these staggering trends
sums, the tangible benefits remain elusive. Adding such
to the tempered outlook, MIT economist and Nobel
as Agentic AI and synthetic data
laureate Daron Acemoglu provides a cautious
evaluation of AI’s economic impact. His research generation expanded AI’s capabilities
suggests that contrary to ambitious forecasts of by enabling autonomous, multi-step
transformative productivity gains, AI may yield
tasks and addressing data scarcity
GDP growth of a more modest 0.93% to 1.16% over
the next decade, with the possibility of reaching
1.56% under optimal conditions. These critiques
underscore the need to balance enthusiasm with
realism, tempering grand visions with practical reasoning and accuracy. Landmark achievements,
assessments of AI’s current capabilities and including Nobel-recognized contributions to
its path forward. protein structure prediction (AlphaFold2) and
industry-specific LLMs for domains like healthcare
and finance, highlighted the technology’s potential.
The year of exponential Global investment in GenAI surged, driven by
tech giants like Google, OpenAI and Microsoft.
breakthroughs Record-breaking funding rounds and open-source
contributions from Meta and others intensified
Year 2024 proved to be one of phenomenal competition, while advancements in hardware,
advancement in the field of GenAI culminating such as Nvidia’s Blackwell platform, provided the
with the announcement of OpenAI’s o3 class of computational power to support increasingly
models, which promise to offer a quantum leap in sophisticated models. Emerging trends, such
foundational LLM capabilities and reasoning. Earlier as Agentic AI and synthetic data generation,
in the year, the transition to multi-modality allowed expanded AI’s capabilities, enabling autonomous,
seamless handling of diverse data formats, while multi-step tasks and addressing data scarcity.
advancements like expanded context windows and SLMs offered cost-effective solutions for smaller
retrieval-augmented generation (RAG) improved enterprises. Despite concerns like overfitting
and model collapse, GenAI’s strides in reasoning,
multimodality and adaptability cemented its
position as a key driver of innovation and
productivity across sectors.

Every once in a while, We are still early in the game

a new technology, an Despite challenges, even today’s innovations in

GenAI offer immense enterprise value. The focus
old problem, and a is not just on GenAI but also on integrating AI,
data, and automation to build tailored solutions.
big idea turn into an Rapid advancements have made AI ‘good enough’
for scaling across many use cases. Techniques like

innovation RAG and CoT address issues like hallucination,

while guardrails secure data privacy and safety. The
cost of AI has also dropped significantly, promising
Dean Kamen returns on existing investments.
Engineer and entrepreneur

The AIdea of India: 2025 17

Human-like adaptability of AI Agents
AI Agents operate, within an enterprise context, to achieve specific goals. They can be instructed in natural
language and act autonomously on behalf of users. Users specify objectives in terms of ‘what’ or task
goals, leaving the AI agent to figure out ‘how’ this is to be accomplished using available tools. An agentic
architecture represents a fundamentally new approach to building computer systems. If successful, it
signifies a leap forward as the focus is on outcomes rather than processes.
A key innovation is that much of the control logic in an AI Agent is driven by LLMs. This approach introduces
dynamic, non-deterministic behavior – similar to human decision-making – with its associated benefits and
challenges. Decision making, with AI agents, is no longer limited to rigid programming. Agents can adapt to
contexts and improve outcomes dynamically.

Applications of AI agents across contexts

Personal assistants Reasoning Agents

Advanced personal assistants, such as Apple’s OpenAI’s O1 (and now O3) models exemplify
AI-driven assistant, showcase how AI can AI’s growing reasoning capabilities. Using
handle complex, context-dependent queries. chain-of-thought methodology, O1 formulates
For instance, when asked about a family step-by-step plans to solve problems, improving
member’s flight arrival and dinner plans, the both accuracy and transparency. Users, too,
assistant seamlessly integrates information can trace the model’s logic, identify errors and
from emails, messages, maps, calendars and make corrections. Notably, O1 has achieved over
third-party apps. These systems build a semantic 80% accuracy in solving complex mathematical
model of the user, which enables navigation problems, marking a substantial advancement
across applications to respond accurately. As over previous models. Reasoning agents
AI becomes more embedded in devices and highlight the potential for AI to bring clarity and
productivity tools, personal assistants are poised reliability to intricate problem-solving tasks.
to adeptly manage digital lives, streamlining user
interactions and enhancing productivity.

Functional Agents Agents in the real world

Salesforce’s Agentforce platform brings Anthropic’s research on AI Agents emphasizes
agentic architectures into the enterprise realm. their ability to interact dynamically with the
These autonomous AI Agents personalize world to accomplish tasks and learn from
customer interactions, streamline support and those interactions. This vision extends beyond
orchestrate actions across multiple channels. the digital realm to where agents can control
This innovation shifts traditional business physical tools, robots or laboratory equipment,
models toward outcome-based pricing – where or even design equipment for specific tasks.
costs are tied to completed tasks rather than However, such dynamic systems bring
per-user licenses. Such a model aligns software challenges in ensuring safety, reliability, and
costs more closely with business outcomes, predictability – an essential focus
offering enterprises a flexible and value-driven for developers.
approach.
18 The AIdea of India: 2025
However, adoption remains low. Our survey of
Indian enterprises suggests that 36% of enterprises 02 Open source LLMs
have budgeted and started investing in GenAI The emergence of open source LLMs (OS LLMs)
while another 24% are experimenting with it. from organizations like Meta and Mistral intensified
Technology sector clients have been leading the competition, prompting closed-source providers
way with Life Sciences and Financial Services such as Anthropic and OpenAI to enhance
following suit. At the same time business value their offerings to justify premium pricing. For
delivered is relatively low with only 15% of Indian instance, DeepSeek v3 has been able to surpass
enterprises report having GenAI workloads in OpenAI’s GPT-4o in performance across several
production, and just 8% being able to fully measure industry benchmarks.
and allocate AI costs. The survey highlights the
need for packaged solutions to bridge the gap and The shift also benefited hardware providers like
accelerate adoption. As innovations mature, they NVIDIA. Demand for GPUs expanded to include
will drive a new wave of digital transformation, organizations deploying OS LLMs privately, leading
unlocking extraordinary business benefits. At them to invest in NVIDIA hardware to run models
the same time, global trends positively influence like Meta’s Llama 3.1 405B internally,
GenAI developments in India through collaboration, rather than relying on API-based access to
investment and research. closed-source models. This diversification of
customer base potentially stabilized GPU demand,
Key GenAI trends across the globe reducing reliance on a few large-scale buyers.

01 Multimodal AI
Deploying private OS LLMs offered advantages
for scenarios where data sensitivity and privacy
Traditionally, processing varied data forms required were paramount. Industries such as healthcare
separate, specialized models, each tailored to and finance utilized these models within secure
a specific modality. This approach necessitated environments, complying with stringent data
complex pipelines to combine outputs from protection regulations. For example, a financial
different models, often leading to increased institution fine-tuned an OS LLM on proprietary
computational overhead and latency. Recent transaction data to detect fraudulent activities
advancements have led to the development of without exposing sensitive information externally,
unified multimodal models capable of processing enhancing operational efficiency while maintaining
multiple data types within a single framework, robust data privacy standards.
streamlining integration and enhancing efficiency.
The evolution of multimodal models brings AI
systems closer to human-like data processing
03 Use of Agentic AI and tools
capabilities. Further, a significant advantage of The advent of Agentic AI has changed the way
multimodal models is the reduction in latency LLMs are deployed. Unlike traditional LLMs that
when processing mixed-form content. By handling passively generate text based on input prompts,
multiple data types within a single model, these Agentic AI endows these models with the capability
systems eliminate the need for separate processing to autonomously plan, reason and execute tasks.
stages and the associated data transfer. This This evolution enables LLMs to function as
unified approach not only accelerates processing proactive agents, performing complex operations
times but also reduces the computational resources without continuous human intervention. For
required, making it more feasible to deploy instance, OpenAI’s forthcoming autonomous AI
advanced AI capabilities in real-time applications. Agent, code-named “Operator,” is designed to
operate independently, marking a substantial shift

GenAI’s
from passive response models to active,
decision-making entities.
wide impact
The integration of tool-use capabilities further
means the enhances the functionality of LLMs, allowing
future of work requires AI-specific them to interact with external systems and APIs
talent and reskilling. While many to retrieve information, perform calculations, or
execute specific commands. This development has
enterprises use Ai tools, few have the led to the creation of more dynamic and versatile
required in-house AI talent

The AIdea of India: 2025 19

AI applications. Frameworks like LangChain have OpenAI’s GPT-4o, o1 and o3 exemplify these
evolved to incorporate tool use, allowing the advancements, excelling in reasoning tasks
integration of third-party APIs ranging from simple and serving as a cognitive partner for decision-
Google search to niche and domain specific APIs. makers. Their applications are particularly notable
This broadens their applicability across various in machine learning, where they aid with tasks
domains. Consequently, LLM deployment has such as data preprocessing, model training and
evolved from static, text-based applications to more optimization, and performance evaluation.
interactive and autonomous systems capable of
Google has also advanced the field with
complex, multi-step tasks, thereby expanding their
math-focused reasoning models like AlphaProof
utility in real-world scenarios.
and AlphaGeometry 2. These models are designed
AI Agents are a promising step forward and we can to handle intricate scientific reasoning and have
expect a significant portion of software in particular achieved high performance, ranking in the 89th
and the internet at large to be re-written for use by percentile for competitive programming tasks. By
agents rather than humans. generating hidden reasoning tokens to internally
‘think through’ problems before producing final
AI Agents promise to redefine efficiency,
visible answers, these models demonstrate
reasoning and personalization across industries.
significant progress in addressing complex
As their capabilities evolve, these agents will
reasoning challenges.
seamlessly blend digital intelligence with
human-like adaptability, empowering enterprises
and individuals alike.
05 New chips, more powerful devices,
new form factors
04 Reasoning in foundation models NVIDIA dominates the hardware market with a
The question “Can LLMs reason?” became one of 70% to 90% market share. Its Blackwell
the most debated topics of 2024. While skeptics microarchitecture significantly boosts GenAI
like Nobel laureate Geoffrey Hinton and Professor capabilities, supporting LLMs with trillions of
Subbarao Kambhampati questioned their reasoning parameters and offering up to 25 times the
capabilities, others, such as former OpenAI chief energy efficiency of prior models. Additionally,
scientist and founder Safe SuperIntelligence Inc. the Nemotron-4 340B model suite produces
Ilya Sutskever, defended the notion that LLMs high-quality synthetic datasets, which are crucial
possess some form of reasoning ability. However, for training robust LLMs in various sectors.
a consensus is emerging across the industry Apart from traditional competitors Intel, Huawei
that, in their current state, LLMs are unlikely to and AMD, Amazon, Meta, Google and start-ups
achieve artificial general intelligence (AGI), though such as Cerebras Systems, Groq, and SambaNova
this could change based on early independent Systems are also pushing the boundaries in this
assessments of OpenAI’s o3 models. field as AI accelerators and GPUs are becoming
CoT prompting has become the leading technique increasingly vital to the industry’s future.
to enhance LLMs’ ability to tackle complex
problems such as in domains like science,
programming and mathematics.

20 The AIdea of India: 2025

The global chip race

Company Chip Focus area

Established companies

NVIDIA Blackwell B200 Next-generation GPU architecture for AI and deep learning

AI processing for data centers; competes with

AMD MI325X
NVIDIA’s offerings

Intel Gaudi 3 AI training and inference processors for data centers

Custom AI inference accelerator for deep learning,

Meta MTIA
optimized for recommendation models

AI performance advancements for mobile devices,

Qualcomm Snapdragon 8 Gen 3
targeting edge AI applications in smartphones

Custom chips

AI accelerator for cloud workloads, LLM training, ARM-

Microsoft Azure Maia 100, Cobalt 100
based CPU for general-purpose cloud workloads

Enhanced performance for AI workloads, part of Google

Google TPU v6
Cloud infrastructure

AWS (Amazon) Trainium 3 Custom chip for training machine learning models

Start-ups

Wafer-Scale Engine World’s largest chip, designed to accelerate deep learning

Cerebras
( WSE-2) models across data centers

High-performance RISC-V CPU as a cost-effective

Rivos Custom RISC-V CPU
alternative to NVIDIA’s GPUs

LPU Optimized for LLMs, enhancing speed and efficiency

Groq
(Latency Processing Unit) in AI tasks

List is not exhaustive

06 New device form factors redefine

the user experience
Samsung’s on-device AI processing, and Deutsche
Telekom’s app-less T-Phone highlight the trend
toward more seamless, context-aware interfaces
GenAI, multimodal LLMs, and specialized chips Emerging technologies like smart glasses
are leading to changes in device design, which is and compact assistants further redefine user
moving beyond conventional smartphones and experiences. Meta’s smart glasses integrate
laptops. AI-integrated devices like ‘AI phones’ AI-powered voice controls, while Rabbit Inc.’s R1
are emerging, offering highly personalized prioritizes AI Agents over traditional apps.
services tailored to user behavior. These devices Start-ups are creating edge-focused chips for
feature advanced AI-powered assistants that can devices, enabling local AI inference for faster, more
perform tasks such as email summarization and efficient processing while reducing server reliance.
recommendations. Apple’s advanced writing tools,

The AIdea of India: 2025 21

With rising operational costs and issues such as Companies can thus deploy sophisticated AI
energy supply, energy-efficient designs are critical capabilities without incurring the high costs
for scaling AI solutions. The industry is prioritizing associated with running large-scale models,
lower power consumption without sacrificing making AI adoption viable for smaller
performance, paving the way for broader adoption enterprises.
of live AI inference across applications.
Domain expertise and precision

07 Small language models (SLMs) SLMs excel in tasks requiring deep domain
knowledge. Whether it is legal document
analysis, financial forecasting, or medical
LLMs offer remarkable capabilities but often come diagnostics, these models are fine-tuned
with high computational costs, making them less to understand the specific language,
accessible for smaller businesses. In contrast, terminologies, and nuances of a given
SLMs provide a more cost-effective alternative. field. This enhances their precision and
Optimized for specific tasks, SLMs leverage relevance, outperforming generalized
fine-tuning techniques to reduce computational models in specialized tasks.
demands, enabling them to run efficiently on edge
devices. Leading examples include Microsoft’s Data privacy and security
Phi-4, DeepSeek v3, Google’s Gemma, By deploying SLMs at the edge, businesses
Alibaba’s Qwen 2.5, Meta’s Llama 3.2, can minimize the need to transfer sensitive
and Mistral’s suite of SLMs. data to centralized servers. This reduces
SLMs combine robustness, safety, and adaptability, privacy risks and ensures compliance with
making them ideal for low-latency, limited-scope stringent data protection regulations,
applications. They can also be securely deployed a critical factor in industries such as
on-premise or in private cloud environments, healthcare, finance, and government.
significantly reducing the risk of data leakage.
Scalability and collaboration
As practical and efficient tools, SLMs are Domain-specific SLMs enable organizations
well aligned with the needs of modern businesses to scale AI initiatives across diverse use
and are poised to play a transformative role in the cases efficiently. For instance, a retail
future of enterprise AI solutions. company could deploy one SLM for
inventory management and another for
The role of SLMs in the enterprise personalized marketing. Open collaboration
SLMs have been a powerful tool in the arsenal of within AI ecosystems, powered by platforms
enterprise architect for a number of reasons. like Hugging Face, facilitates sharing pre-
trained domain-specific models, accelerating
Edge deployment innovation while reducing time-to-market.
SLMs are lightweight and optimized
for deployment at the edge—closer to As domain-specific SLMs gain traction, they will
where data is generated and consumed. play an increasingly pivotal role in reshaping
This is particularly valuable in scenarios enterprise applications. By integrating edge-ready,
requiring low-latency responses, such as in cost-efficient, and highly specialized models,
manufacturing, healthcare devices, or IoT businesses can unlock the full potential of AI while
systems. For instance, an SLM trained for ensuring sustainable and scalable deployment.
predictive maintenance can operate on-site
in a factory, analyzing machine data in real-
time without relying on cloud connectivity.
08 Continued
technology
investments into GenAI

Cost efficiency
Unlike LLMs, which demand significant While investment in GenAI has been on the rise
computational resources for both training since 2018, 2023 was a breakout year and this
and inference, SLMs are designed to momentum continued in 2024. The funding surge,
operate on smaller-scale infrastructure. particularly from giants like Microsoft, Amazon and
This leads to lower operational costs, Google, is reshaping the business landscape and
including reduced energy consumption and leading to transformative changes. While OpenAI
the need for fewer hardware resources. raised US$6.6 billion in October 2024, boosting its

22 The AIdea of India: 2025

Large investments driving GenAI innovation for enterprise use

Company Amount Investors Announced Funds to be deployed towards

Institutional Venture
Perplexity US$500 Enhancing AI-powered search
Partners, SoftBank, Dec-2024
AI million capabilities and expanding market reach
Nvidia, Jeff Bezos

Thrive Capital, Insight

US$10 Expanding cloud-based AI technology
Databricks Partners, Nvidia, Dec-24
billion solutions and preparing for potential IPO
Andreessen Horowitz

Enhancing AI-driven healthcare

AI Fund (founded by
Jivi Undisclosed Oct-24 solutions, including diagnosis and
Andrew Ng)
treatment recommendations

NTTVC, Z47 (formerly Scaling AI infrastructure, advancing

US$30
Neysa Matrix Partners India), Oct-24 R&D, and launching integrated Gen AI
million
Nexus Venture Partners Acceleration Cloud Service

Thrive Capital, Softbank To enhance AI research and increase

US$6.6
OpenAI Vision Fund 2, Coatue Oct-24 computing capacity; valued at
billion
and others US$157 billion

US$4 billion
Anthropic Amazon Sept-24 Develop AI models on AWS infrastructure
follow on

US$41 Developing large language models

Sarvam AI Not publicly disclosed Jul-24
million focusing on Indian languages

US$1.1 Coatue Capital, Scale operations for cloud services

CoreWeave May-24
billion Magnetar and others tailored for AI workloads

For product development and

xAI US$6 billion Various VCs May-24
infrastructure expansion

Moonshot US$1.3 Alibaba Group, Tencent, Advancing AI technologies with a focus

Feb/Aug-24
AI billion Gaorong Capital on long-context language models

Jeff Bezos, Microsoft,

US$675 Nvidia, Intel, Amazon’s Developing AI-powered humanoid robots
Figure AI Feb-24
million and OpenAI’s start-up for industrial applications
funding divisions

Major investments of 2024 included; list is not exhaustive

valuation to US$157 billion, Amazon invested in in 2024, reshaping industries and redefining
Anthropic with a portion of a US$4 billion round, possibilities. While these developments are
reinforcing its commitment to advancing GenAI universal, their impact often takes unique forms
technologies on its cloud platform. The company in different regions. In the Indian context, where a
continues to invest heavily in infrastructure that dynamic mix of challenges and opportunities exists,
supports GenAI applications, partnering with it becomes imperative to identify the critical forces
various start-ups to enhance capabilities. Further, that will drive disruption. The following section
NVidia and Google’s Deepmind are funding several delves into five key vectors of disruption that are
specific projects aimed at advancing R&D in AI. poised to shape India’s GenAI journey, spotlighting
the areas where this transformative technology is
The rapid evolution of the GenAI landscape has
expected to make the most significant impact.
brought transformative changes across the world

The AIdea of India: 2025 23

24 The AIdea of India: 2025
GenAI in India: This is already beginning to happen. More than one-
third of Google searches in India are voice-based, a
Key trends stark contrast to just 5% in developed markets. This
growing preference for voice interaction is evident
India’s GenAI landscape is rapidly evolving, in initiatives like the AskDisha chatbot by CoRover,
characterized by unique market dynamics and which supports ticket bookings on the IRCTC app
a burgeoning ecosystem of innovation. Unlike through voice, chat, and point-and-click interfaces.
developed markets, India’s GenAI trajectory is Similarly, the National Payments Corporation of
shaped by its vast and diverse population, a India (NPCI) has pioneered solutions such as Hello!
young and tech-savvy workforce, and specific UPI, which enables transactions via simple voice
socio-economic challenges. In this chapter, we commands. Feature phone users and those in low-
explore five key vectors driving GenAI disruption connectivity areas can use UPI 123PAY through
in India, highlighting both the tailored adoption of voice prompts, missed call services, and callback
global solutions and the emergence of indigenous mechanisms, all without an active
products and solutions designed to meet internet connection.
local needs. The new interfaces also have the potential to
transform accessibility for underserved sections
01 Chat, voice, regional languages
augment digital interfaces of India, particularly in semi-urban and rural
areas. There is a need for affordable solutions in
Using mobile apps can be hard work. Screens are regional languages, intuitive interfaces and services
small, the navigation even for simple transactions designed for specific needs. GenAI is emerging as
can become complex and it is not easy to a transformative force, enabling innovations that
personalize the experience. For the less digitally drive financial inclusion, healthcare access, and
savvy and those more comfortable in regional educational outreach through localized,
languages, apps and by extension a large part of cost-effective solutions.
the digital economy are largely out of reach.
In education, AI-powered platforms will enable
GenAI is driving a dramatic change in digital personalized tutoring in regional languages,
interfaces. One can see the emergence of a new UI offering adaptive learning experiences tailored
— led by AI assistants, activated by voice, extended to individual progress and needs. In healthcare,
into regional languages and augmented by chat. GenAI will address resource shortages by enabling
This will rapidly begin to augment and in some remote diagnostics, analyzing longitudinal
cases replace the current app-based point and click health data, and delivering personalized care
model. Over time, one can even imagine an LLM recommendations. These advancements promise
dynamically generating the UI based on the context to redefine accessibility, empowering communities
of the transaction. and creating a more inclusive digital ecosystem
across India.

India’s consumer landscape, unique characteristics and market dynamics

India 1 India 2 India 3

Comprises the top 10%-15% of Encompasses approximately The largest segment, with
households and approximately 300 million individuals from approximately 900 million
100 million people. This smaller cities and semi-urban people, is predominantly
segment is affluent, urban, areas. These consumers rural, lower-income, and
and digitally savvy, driving have moderate purchasing less digitally literate,
a significant portion of the power, are price-sensitive yet communicating mainly in
digital economy and consuming aspirational, and seek regional languages and
premium products value-driven services in focusing on affordable,
and services. education, healthcare and essential services.
finance.

The AIdea of India: 2025 25

How business strategies could differ for India 1 and India 3
Industry Digital model India 1 strategy India 3 strategy
Personalized investment management, AI-based micro-lending, regional
AI-driven wealth advisory, predictive language support in financial apps,
Fintech analytics for portfolio performance voice-based banking interfaces

Predictive risk assessment, Affordable micro-insurance through

Insurtech personalized insurance policies, AI-based risk assessment, simplified
Financial automated claims processing AI chatbots for policy inquiries
services Credit scoring
AI-driven credit scoring models Credit scoring for the unbanked
using alternative data (social, using local data (e.g., mobile
transaction history) transaction patterns), low-cost
loans accessible via mobile

Personalized content curation, AI- Regional language content

driven recommendation engines, recommendations, AI for local news
targeted advertising based on curation, audio/voice-based access
Digital media behavior analysis to content

Entertainment AI-generated media content Regionalized AI-driven video

Media (e.g., news articles, video recommendations, low-data
recommendations), immersive AR/ streaming options, automated
VR experiences dubbing or subtitles

AI-driven diagnostics, predictive AI-based mobile health diagnostics,

offline AI health consultations,
health analytics, telemedicine with
regional language support for
AI-driven triage healthcare apps
Healthtech
AI-driven virtual consultations Mobile-based telemedicine with
Telemedicine basic AI for symptom checks, local
with specialists, medical imaging
language support for rural health
Healthcare analysis workers
Pharmaceuticals
AI-led drug discovery, personalized Low-cost AI solutions for public
health issues, predictive analytics
medicine, genetic analysis for
for vaccine distribution in
tailored treatments rural areas

AI-driven personalized shopping AI-driven local language shopping

experiences, dynamic pricing, apps, voice shopping, inventory
targeted product recommendations optimization for local vendors
E-commerce
AI-based demand forecasting, real- Simplified AI tools for small
Supply chain retailers, inventory management
time supply chain tracking, and
for local shop owners, predictive
Retail and logistics optimization demand analytics
E-commerce Payments
AI-powered fraud detection, AI-based digital payments using
local authentication methods,
seamless digital payments,
voice-activated transactions in local
biometric payments languages

Personalized learning paths, AI-driven Localized, AI-driven learning

content in regional languages, basic
tutoring, content recommendations
mobile AI tutoring, voice-activated
based on learning patterns education platforms
Edtech

AI-based proctoring, adaptive Low-cost, AI-based grading systems

Assessments for large groups, offline AI tools for
testing, automated grading
Education grading and assessments
Skill training AI-powered skill training for local
AI-driven upskilling and reskilling
workforce in regional language,
programs with personalized
mobile-based skill assessments and
learning tracks recommendations

26 The AIdea of India: 2025

02 Agents
Integrating
enable the
transformation of knowledge
work
GenAI with traditional methods,
The advent of AI Agents — autonomous
LLM-powered software capable of understanding compound AI systems are leading to
context, accessing a toolkit of resources skillfully, more robust, efficient and intelligent
making decisions, and executing actions iteratively
applications across various sectors that
towards achievement of a goal — and agentic
orchestration promises to further enhance harness these combined strengths
advancements. The technical aspects of agents is
discussed in detail in the previous section. In customer support, AI-driven chatbots and
virtual assistants efficiently handle routine
India’s significant role in the global knowledge inquiries, providing 24/7 support and prompt
economy is evident through its extensive workforce resolutions. AI tools today promise to manage a
in sectors like information technology, finance spectrum of customer interactions, from account
and customer service. The rapid integration of AI balances and transaction histories to more complex
Agents into these fields is reshaping traditional financial advice, thereby enhancing customer
ways of working, presenting both opportunities satisfaction and operational efficiency.
and challenges for Indian professionals. Most
notably, Indian knowledge workers have leaned in, By automating routine tasks, AI Agents will enable
with 92% of them utilizing AI in their workplaces, professionals to focus on strategic, creative, and
meaningfully surpassing the global average of complex aspects of their work, thereby enhancing
75%, according to the 2024 Work Trend Index by productivity and fostering innovation. For India,
Microsoft and LinkedIn. with its extensive knowledge workforce, embracing
AI Agents and agentic architectures is crucial to
In software development, AI-powered code editors maintaining competitiveness and harnessing the
such as Cursor can comprehend code context, full potential of its human capital. The impact of AI
suggest bug fixes, and implement related changes on human effort reduction and elimination extends
across entire files. Features like inline chat enable beyond knowledge work, as detailed in the chapter
developers to direct the AI for modifications ‘Transforming work with GenAI’.
ranging from minor edits to extensive codebase
overhauls. This evolution allows programmers to
focus more on system architecture and logic, as AI
manages repetitive tasks. 03 LLMs are not all you need:
Toward compound AI systems
The legal sector is experiencing a transformation
through AI in contract drafting and document It is becoming increasingly clear that LLMs have
review. Agentic tools like Type integrate with legal significant limitations. They are expensive to use,
databases and precedent libraries, expediting hallucinate and often miss out on key context that
the drafting process. These AI solutions enhance is critical for enterprises as they seek to build
accuracy by suggesting pertinent clauses and intelligent systems.
ensuring compliance with regulatory standards.
Enterprises will increasingly rely on hybrid
Financial analysts benefit from AI systems that architectures, where multiple AI models — ranging
automate report generation, document processing, from small, task-specific models on edge devices
compliance monitoring, and anomaly detection. AI to large foundational models in the cloud — are
streamlines processes such as invoice management orchestrated to work together. These models will
by extracting and integrating data into ERP operate at different levels of the enterprise stack,
systems, reducing manual errors. balancing cost, latency, and performance needs.
For example, lightweight models might handle
Content creators leverage GenAI tools capable of
quick, edge-based tasks, while more complex
producing drafts, performing grammar and style
models, fine-tuned to enterprise data, will reside in
checks, and offering SEO optimization suggestions.
cloud environments.
These platforms enable writers and editors to focus
on creative ideation and storytelling. AI-powered In February 2024, the Berkeley AI Research (BAIR)
systems analyze extensive data to generate Lab introduced the concept of compound AI
personalized content, enhancing engagement systems, which integrate LLMs with traditional
and relevance.

The AIdea of India: 2025 27

How GenAI models amplify overall system capabilities

Construction Recommendation
Financial Industry
project engines in online Education
services 4.0 + GenAI
management commerce
Advanced project In e-commerce, Loan applicants Adaptive testing GenAI combines
management images — such interact with has evolved traditional ML
software now as those of a GenAI-powered with GenAI, models with
incorporates living room — are chatbots to extending beyond diverse inputs
drone footage analyzed using provide detailed multiple-choice like workers’
to assess multi-modal vision information and to include long- logs and product
construction sites. models. The answer follow-up form subjective manuals. This
GenAI processes interpretations questions. The answers. This fusion enhances
this visual data, inform classical collected data advancement predictive
providing insights ML-based is processed by allows for more maintenance
that integrate recommendation traditional ML comprehensive strategies, leading
with traditional engines, models, refining assessments, to more informed
AI models to suggesting optimal credit decision- tailoring question decisions on the
enhance project décor options making processes difficulty based factory floor
oversight and from catalogs. and improving on previous and reducing
decision-making. This integration accuracy in risk responses downtime.
improves the assessment. and providing
relevance and The ensemble a deeper
personalization of significantly understanding
recommendations. reduces time to of student
decision. capabilities.

AI/ML techniques, classical programming, and doctor, including tailored product suggestions,
external tools like internet search APIs. This comparisons to regional peers, and relevant
approach marks a big shift from relying solely disease trends.
on monolithic models to employing multifaceted
The evolution of Compound AI systems signifies
systems that leverage the strengths of various
a transformative approach in AI, where the
components.
integration of GenAI with traditional methods
GenAI exemplifies this paradigm by enhancing leads to more robust, efficient, and intelligent
existing systems across diverse industries. Rather applications across various sectors. As this trend
than serving as standalone solutions, GenAI models continues, we anticipate the development of
are increasingly embedded within traditional AI innovative products that harness the combined
and ML processes, creating synergistic effects that strengths of these technologies, driving significant
amplify overall system capabilities. advancements in their respective fields.

Multimodal
Compound AI systems are already prevalent today.
A leading Indian pharmaceutical company relies
on its field sales force to engage doctors and
share product updates. Leveraging their extensive
datasets such as visit logs, doctor feedback, models offer significant advantage of
prescription patterns, and local disease insights,
which is a mix of quantitative and qualitative reduced latency when processing mixed-
data. The revamped sales app now leverages form content. The faster processing times
all of these through traditional AI and GenAI to and less need of computational resources
deliver personalized recommendations for each
enhance real-time application

28 The AIdea of India: 2025

 04 The falling cost of AI are further reducing hardware costs. Combined
with affordable cloud services, these innovations
are empowering businesses to scale AI initiatives
The cost of implementing AI solutions has
and push the boundaries of innovation, driving
plummeted, making them increasingly accessible to
unprecedented growth in AI applications.
enterprises. OpenAI’s GPT API costs, for example,
have dropped nearly 80% in two years, while India’s traditionally low labor costs have led some
open-source releases like Meta’s Llama are to question the cost versus quality trade-off in
unlocking new capabilities. India-specific deploying LLMs to augment human effort in the
fine-tuned versions of these models enable Indian market. However, the rapidly decreasing
affordable customization. Although GPU supply costs associated with AI deployment are reshaping
remains tight, alternatives are emerging, allowing this perspective.
AI inference to run without high-end GPUs.
In August 2024, founder of DeepLearning AI and
Open-source models such as Meta’s Llama series, educator Andrew Ng noted, “Let’s say you build an
Mistral, Google’s Gemini, Microsoft’s Phi, and application to assist a human worker, and it uses
Alibaba’s Qwen are providing powerful tools 100 tokens per second continuously:
at a fraction of the cost of proprietary models. At $4 per million tokens, you’d be spending only
Innovations like prompt caching and batch $1.44 (~INR120) per hour – which is significantly
processing further enhance affordability for LLM lower than the minimum wage in the U.S. and
utilization. many other countries.” With Sarvam AI’s offer of
INR1 per minute, this cost drops further to half as
Cloud providers are also driving cost reductions
much (INR60 per hour), making LLM-led offerings
with pricing based on Tensor Processing Units
extremely price competitive in the Indian context.
(TPUs), delivering faster and cheaper AI solutions.
Improved GPU access has enabled organizations to The convergence of declining AI deployment costs
fine-tune and deploy open-source models, reducing and a deep pool of tech talent capable of building
reliance on heavy infrastructure. Hybrid strategies, AI applications presents a compelling opportunity
combining on-premise solutions for sensitive data for widespread AI adoption. As AI becomes more
with cloud-based APIs for scalability, are proving to affordable, even for small and medium-sized
be cost-effective to expand AI deployments. enterprises, it has the potential to drive innovation,
enhance productivity, and contribute to economic
Technological advancements like quantization,
growth across various sectors.
which optimizes AI models for CPU efficiency,

Token cost of GPT-4 level models over time

Cost for 2 million tokens (input + output) decreased from
US$180 to US$0.75 in two years, becoming 240x cheaper

250

200 gpt-4-32k
$/million tokens

150

gpt-4
100
gpt-4-turbo
50 gpt-4o gpt-4o-mini

0
4/1/23 7/1/23 10/1/23 1/1/24 4/1/24 7/1/24

Source: Elad Gil on X

The AIdea of India: 2025 29

05 The evolution of an Indic AI
ecosystem
Abu Dhabi-based G42 introduced Nanda, a Hindi
LLM trained on 2.13 trillion tokens, aiming to
empower over half a billion Hindi speakers with
GenAI capabilities. Similarly, BharatGen, the first
A lot has been said about India being the use case
government-funded multimodal LLM initiative,
and data capital of the world. With more than
focuses on creating efficient and inclusive AI
700 million connected consumers and the lowest
tailored to Indian needs. Tech Mahindra’s Project
data rates in the world, India will play home to
Indus further advances the development of Indic
many sunrise industries that will seek to infuse AI
foundational models, beginning with Hindi and
at scale across their business models. There is also
its over 37 dialects, bridging linguistic gaps for
now a significant focus on integrating AI into India
enterprises. TWO.ai, another notable player,
Stack to buttress its already strong digital public
has introduced SUTRA, a multilingual AI engine
infrastructure. Will India see an Indic AI stack with
supporting over 50 languages, including Hindi and
tools that compete with global offerings and is
Gujarati. SUTRA is designed to power immersive
relevant in an enterprise context?
AI experiences that extend beyond text and voice,
addressing India’s linguistic diversity effectively.
The race to build Indic LLMs and agents
AI4Bharat, a research lab at IIT Madras, has
One significant area of focus has been the been instrumental in advancing Indian language
linguistic diversity of India — there has been a technology. Their contributions include extensive
mushrooming of Indic LLMs that leverage open-source datasets like IndicCorp v2 and tools
open-source models fine-tuned with Indian that underpin the growing Indic LLM ecosystem.
language datasets. A key initiative in this space Despite these advancements, challenges persist.
is Bhashini, a government-led AI project aimed at The lack of robust datasets across India’s diverse
creating an open-source Indic language dataset to languages affects the training and performance
expand internet and digital service accessibility in of these models. Data often requires significant
Indian languages. By facilitating content creation cleaning and processing to ensure usability, while
in languages like Hindi, Tamil and Bengali, Bhashini concerns about responsible data use, privacy, and
is democratizing the benefits of AI for India’s ethics remain pressing. Initiatives such as Project
multilingual population. Vaani, a collaboration between IISc and Google
to collect Indic speech datasets, are critical in
The global AI community has also expressed keen
addressing these gaps.
interest in Indic languages. For example,

30 The AIdea of India: 2025

The development of AI Agents tailored to Indian 75% generating revenue in the first half of 2024
contexts represents another promising area. compared to just 22% in the same period the
These agents not only address linguistic diversity previous year. The ecosystem now includes 17
but are fine-tuned with domain-specific datasets, native GenAI language models and a significant
enhancing their utility in sectors like BFSI, increase in start-ups offering GenAI assistants,
healthcare, and legal technology. Sarvam AI leads placing India sixth globally in GenAI start-up
the way with its GenAI platform featuring ecosystems among major economies.
Sarvam Agents — voice-enabled, multilingual
solutions designed for Indian enterprises. Their
open-source models, such as Shuka 1.0 Towards a sovereign AI cloud
(an open-source AudioLM), and APIs supporting
In parallel, the development of digital infrastructure
10 Indic languages, mark significant milestones.
in India has gained momentum. Leading
Sarvam AI’s cost-effective pricing model of
corporations such as Reliance, Tata Group, Tech
INR1 per minute for AI-driven customer
Mahindra, and Wipro have partnered with Nvidia
interactions ensures accessibility for businesses
to procure advanced GPU infrastructure. Mid-sized
of all sizes. Similarly, Kogo.ai has developed a
players like Yotta Data Services, E2E Networks,
platform enabling companies to build AI agents
and Sify are also providing cloud GPU resources to
that can converse in Indic languages. Starting with
enable the hosting of LLMs. Moreover, the Ministry
Hindi, Urdu and English, Kogo.ai plans to expand
of Electronics and Information Technology (MeitY)
its offerings to additional languages, providing
is actively working to empanel partners capable of
versatile solutions for enterprises
delivering up to 10,000 GPUs, thus empowering
seeking multilingual AI capabilities.
start-ups, researchers, and academicians to
Benchmarks play a crucial role in measuring contribute to AI innovation.
advancements in LLMs and Indic LLMs are no
The development of an Indic AI ecosystem reflects
different. IndicGenBench, introduced in 2024, is a
the potential for India to leverage its unique
comprehensive benchmark designed to assess the
linguistic and cultural diversity in shaping AI
performance of LLMs across 29 Indic languages.
innovations. Initiatives like Bhashini, BharatGen,
It includes tasks like cross-lingual summarization,
and the commitment to purchase 10,000 GPUs,
machine translation, and question answering,
along with the rise of GenAI start-ups, demonstrate
providing multi-way parallel evaluation data
significant early progress. However, addressing
for under-represented Indic languages. Such
challenges such as data quality, infrastructure
benchmarks are essential for driving innovation and
gaps, and ethical considerations will be critical for
ensuring the models meet the diverse needs
sustained growth.
of India’s population.
Despite the hurdles, the strides made so far
These developments collectively represent a
underscore a strong foundation for further
growing ecosystem of Indic LLMs and agents
development. Collaborative efforts among
designed to cater to India’s linguistic and cultural
government, industry, and academia are setting the
diversity. From addressing challenges in data
stage for a comprehensive AI ecosystem tailored
quality to building sophisticated AI Agents and
to India’s needs. While there is still much work
creating robust benchmarks, the Indic AI landscape
ahead, the momentum and focus evident today are
exemplifies the transformative potential of AI to
encouraging indicators that India is well-positioned
make technology more inclusive and impactful for a
to create an AI ecosystem that is inclusive,
multilingual society.
impactful, and globally relevant.

A burgeoning GenAI start-up ecosystem

India’s burgeoning GenAI start-up ecosystem has

also contributed to this momentum. The number of
GenAI start-ups in India surged 3.6 times from
over 66 in the first half of 2023 to more than
240 by mid-2024, according to Nasscom’s
“India’s Generative AI Start-up Landscape 2024”
report. These start-ups have cumulatively
attracted over US$750 million in funding, with

The AIdea of India: 2025 31

Chapter 2

Pivoting to AI-first
digital transformation
32 The AIdea of India: 2025
The AIdea of India: 2025 33
Chapter 2

Pivoting to AI-first
digital transformation

S
imilar to the transformative impact of the
digital revolution, the accelerating shift
AI-first strategies are reshaping toward AI-driven platforms is poised to
Indian enterprises by automating reshape every factor influencing a company’s
workflows, optimizing value chains, EBITDA. Across Indian enterprises, AI-first
approaches are steadily taking root, embedding
and introducing new interfaces for themselves throughout the value chain to enhance
enhanced efficiency and value creation operational efficiency and unlock new avenues of
value creation.
Enterprises are adopting domain- A fully reimagined technology stack is
specific models, intelligent automation, emerging — one that enables innovative customer
experiences, drives agentic automation, and
and AI-embedded tools from various prioritizes data as the cornerstone for fueling AI
providers to accelerate deployment and models. This evolution is prompting organizations
enhance operational adaptability to reassess their enterprise risk posture, develop
new frameworks for managing AI in production,
and address evolving cybersecurity challenges.
Robust data governance, scalable
As AI reshapes the nature of work, organizations
modern data stacks and proprietary are redefining role architectures and reimagining
datasets are critical enablers for AI learning methodologies to equip their workforce for
success this paradigm shift.

Change management is essential for Survey: From experiment to

ensuring workforce adaptability and adoption
unlocking the full potential of AI,
bridging the gap between technological We conducted a survey of more than 125 Indian
innovation and practical execution enterprise leaders to understand their experience
with GenAI and their current thinking on it. While
there continues to be significant optimism about
Organizations are addressing risks like GenAI, survey results show a mixed picture when
bias and cybersecurity by leveraging it comes to adoption. While nearly all Indian
explainable AI models, automated enterprises are exposed to GenAI through its
integration into existing software, only 15% of
compliance systems and regulatory surveyed enterprises have directly implemented
frameworks like India’s DPDP Act GenAI in production. A significant 36% have yet
to commence any experimentation, reflecting

34 The AIdea of India: 2025

The agenda for enterprises

Reimagine the Rethink the tech Move to AI-ready Getting your Confronting the
business and stack data people ready for AI changing frontier of
operating model risk
■ Move to an ■ Nurture ■ Build a GenAI
■ Optimize the AI-first app datasets strategy around ■ New risk
existing value strategy: that provide applications for management
chain with AI interfaces, competitive sustainable, frameworks to
models, advantage innovative address the dynamic
■ Enhance and probabilistic
architecture competitive
existing revenue ■ Implement a nature of AI models
gains
streams: ■ Prepare for the robust data
channels, coming wave of governance ■ Develop use ■ Real-time
portfolio, pricing AI-embedded framework ■ cases for governance
enterprise apps productivity, mechanisms (e.g.,
■ Transform ■ Implement a anomaly detection,
automation,
delivery with ■ Leverage next- scalable data explainable AI)
and customer
new interfaces: generation infrastructure ensure transparency
experience
D2C, social, approaches and accountability
digital sales to automation ■ Assess value
(RPA) and realization ■ Better techniques
■ Launch new and training data
integration in revenue,
products and filtering improve AI
savings,
services: for system reliability
efficiency, and
example, India
engagement
3 consumers, ■ AI-powered security
connected ■ Reimagine solutions, cyber
products business defense strategies,
processes and and continuous
workflows learning minimize
risks

AI
the early stage of adoption. Meanwhile, 8% of
enterprises that have experimented with GenAI
-powered automation is slashing
report challenges in realizing tangible impact.
time and costs, transforming
Despite these early struggles, the promise of
GenAI remains undeniable. Its potential to drive
process time from weeks into tasks
efficiency, innovation, and growth across industries completed in hours, setting new
is immense. Indian enterprises stand at the cusp standards for efficiency
of a new era, where those that successfully
harness GenAI will unlock transformative value and
establish a competitive edge in the AI-driven future.

The AIdea of India: 2025 35

Experience with GenAI POC engagements reshape productivity expectations for the Indian
ITeS industry, which may force companies to
Integration with existing software means enterprises’ pass on cost savings to customers and creating a
exposure to GenAI is high. However, only a few have the
competitive environment where such efficiencies
technology in production.
become a baseline.
Beyond cost savings and operational efficiency,
AI is enabling businesses to move toward more
customer-centric models. Companies are
36% 11% reimagining their value chains, launching new
15% products, and creating innovative services.
34%
8% Towards AI-led value chains, products services:
30%
■ Customer service: AI-powered voice agents,
available for just a few rupees per interaction,
POCs completed Productionalization in are reshaping the unit economics of customer
progress
POCs in progress support. Tomorrow’s bank, for instance, could
AI adopted
No POCs done
potentially operate with a fraction of its
No clear impact current workforce.

Source : EY India C-suite GenAI Survey 2024 ■ Consumer products: Companies are
increasingly focusing on D2C
(direct-to-consumer) channels, transforming
search, catalog management, and
Reimagine the business and personalization with AI.
operating model ■ Media: Media organizations are building
AI-driven ecosystems to deliver personalized,
GenAI is poised to drive significant shifts across on-demand content.
business and operating models, reshaping
industries at both foundational and strategic levels.
At a foundational level, AI allows firms to digitize Organization’s readiness in adopting
operations and enhance existing revenue streams
by embedding AI into every aspect of their value
GenAI: Strategy (direction and alignment)
chains. Unlike traditional deterministic automation, More than half of the enterprises have a GenAI strategy
where business logic is hardcoded into applications, but only some have a fully integrated one with clear
AI-powered systems process data streams, detect execution plans
22%
patterns, and generate insights or actions, creating 12%
closed-loop systems for continuous learning.
A GenAI model can analyze various data streams -
conversations over chat or email, a set of dynamic
supply chain variables or employee attrition 18%
data to identify patterns, predict outcomes,
and recommend or automate actions, thereby 9%
fundamentally reengineering business processes.
Consider the IT industry. Amazon CEO Andy
Jassy recently shared how their GenAI assistant
“Q” slashed the time required to upgrade some 39%
apps from 50 developer days to just a few hours,
cumulatively saving 4,500 developer years Fully integrated Strategy aligned with
strategy with clear business goals, but execution
across their upgrade projects. Google reports execution plans plans are lacking
that 25% of its code is now AI-generated, a move Strategy exists, but not Basic understanding,
accompanied by a 25% reduction in its Python aligned with business goals no formal strategy
team over 2024. This automation will dramatically No clear strategy
Source : EY India C-suite GenAI Survey 2024

36 The AIdea of India: 2025

■ Social commerce: Platforms leverage AI to Implications for companies
connect sellers directly with buyers, bypassing
traditional retail models.
Optimize the existing value
■ Agritech: Start-ups are providing AI-powered chain with AI – across customer
farm management solutions, offering engagement, supply chain
data-driven insights to farmers and creating operations, finance and HR
new revenue streams for agribusinesses.
■ Automotive: Auto companies are expanding
Enhance existing revenue
their connected vehicle ecosystems through AI, streams – through advanced
enabling predictive maintenance and enhancing market segmentation, channel
customer experiences. optimization and pricing
GenAI represents more than just an incremental
improvement — it is a transformative force.
By enabling operational efficiencies and
fostering customer-centric innovation, AI will Transform delivery with new
fundamentally reshape industries, creating new interfaces – D2C, social,
digital sales
competitive benchmarks, business models, and
revenue opportunities.

Launch new products and services

– for example, new services
targeting India 3 consumers,
connected products

The AIdea of India: 2025 37

Rethink the tech stack Organization’s readiness in adopting
GenAI: Implementation (buy vs. build
From the GPU chips that power AI server farms approach)
to intelligent agents that are changing the app Approximately one in four have defined approach but
landscape as we know it, a new tech stack is being application is uneven
fashioned.
16% 10%
Beyond LLMs
Companies are using a range of models to meet
their AI needs. This helps balance capability, cost
and data privacy concerns. For instance, companies
in regulated industries like pharmaceuticals use
public LLMs such as OpenAI GPT for non-sensitive
tasks like summarizing content or generating 21%
marketing insights. At the same time, they rely on
private, on-premise open-source LLMs to handle
sensitive R&D data.
21% 32%
The size of a model determines its cost and
SLMs are becoming the workhorses for specific Well-defined and Defined approach, Preliminary
domain bound tasks that do not need the heavy consistently but not consistently approach, not fully
processing power of LLMs. (Read more about applied approach applied defined or executed
the edge deployment, cost, precision, Aware of options, but no clear No consideration
scalability and other features of SLMs in decision framework of buy vs. build yet
‘Generative AI - Shaping tomorrow’.)
Source : EY India C-suite GenAI Survey 2024

Implementation: Building with next generation components

Layer What this means
for enterprises
App
Apps Enterprise-wide experiences
Apps and Build apps with a new
experiences Content Agentic Dashboards and Enterprise knowledge focus on embedded
generation assistants control towers management intelligence

Agent / chain APIs and Create enterprise/group

Apps Intents Infrastructure
Orchestration toolbox orchestration plug-ins and operations wire experience store

Rethink traditional
models of automation
Classic machine Open Proprietary Emerging (RPA, process engines)
ModelsApps learning (ML) models models models
Treat models as
first-class citizens in the
enterprise architecture
Apps Data
Intelligence Embedding Vector Semantic Monitoring and
pipeline models stores indices observability
Enable the move to
modern data stack

Trust model Enforceable guarantees End-to-end encryption

Move to an advanced
Apps
Governance security architecture
Auditing and Security and Regulatory Responsible AI
traceability zero trust compliance and ethics governance model

38 The AIdea of India: 2025

Cost reduction strategies for GenAI implementation
While the cost of intelligence is falling, implementing enterprise-grade AI applications can be complicated.
However, deploying a range of approaches can reduce the total cost of ownership

Strategy Description Impact on costs

Model Development of smaller, more efficient Up to 60% reduction in

optimization models (e.g., GPT-4o mini) operational costs

Open-source Adoption of open-source alternatives (e.g., Lower licensing fees and

models Llama, Gemma, Mistral, Phi and Qwen) customization costs

Infrastructure- Utilizing cloud-based services for Pay-per-use model reduces

as-a-service model training and deployment hardware costs

Fine-tuning Tailoring models for specific tasks to Decreased inference costs

models improve efficiency

Automation of Implementing automated processes to Reduced labor costs and

workflows streamline operations increased efficiency

Data volume Optimizing context to reduce token usage Lower data storage and
management processing costs

Caching LLM responses helps economise

Caching LLM usage and avoid repetitive LLM usage for the Decreased inference costs
responses most common queries/use-cases

Organization’s readiness in adopting The coming tsunami of AI-enabled

GenAI: Architecture (GenAI platform and enterprise apps
integration approach)
Enterprise app vendors are racing to embed AI into
Architecture integration is limited and enterprises are their offerings. This will be the way by which many
looking at ways to increase application enterprises get their first taste of AI. Enterprise
4% 11% apps will seek to use the transaction context
that they are aware of to offer out-of-the-box AI
19%
solutions. This bypasses tricky implementation
decisions related to buy versus build question,
LLM choice and data wrangling – enterprise app
vendors are already incumbents and will have the
advantage of being able to drive quick adoption of
their offerings.
Imagine entering a purchase order into a next
generation ERP. Embedded AI models dynamically
43%
23% analyze historical spend data, supplier performance
and contract terms to recommend optimal
Fully integrated and Integrated architecture in place; purchasing decisions in real time. Fields are
optimized platform facing utilization challenges pre-filled, anomalies are flagged and cost-saving
Platform selected; Platforms identified, but no alternatives recommended, all while ensuring
integration just started integrated architecture compliance with procurement policies. With
No platform selected / integration approach defined
Source : EY India C-suite GenAI Survey 2024

The AIdea of India: 2025 39

seamless integration into spend management For example:
systems, AI enables instant insights into spending
■ ServiceNow’s Now Assist offers incident
patterns, automates approvals, and provides
summarization, code generation, and resolution
analytics to optimize supplier relationships,
note creation, enabling IT agents to expedite
transforming purchase order processing.
tasks and improve efficiency. Recent expansions
Enterprise SaaS behemoths like Salesforce, SAP, include Now Assist in Virtual Agent, which
and Oracle are already racing to integrate AI into facilitates advanced GenAI chat experiences;
their platforms. The primary reasons to choose flow generation, which converts plain text into
app-native GenAI solutions are: low-code workflows for rapid development;
and Now Assist for Field Service Management,
aiding field teams with task summaries to
streamline operations.
01
Seamless integration with the
rest of the application ■ Salesforce’s Einstein delivers conversational
AI across all the organization’s applications,
allowing for proactive suggestions and actions,
such as recommending follow-up tasks,
02 Grounding in app specific data
to create relevance after-sales calls, or generating personalized
marketing content.
■ SAP’s Joule supports tasks like executing
Ensuring privacy and
03 compliance by keeping data
within the application
transactions, navigating applications, retrieving
information, and generating analytical insights.
■ Oracle Fusion brings over 50 GenAI capabilities
directly into business workflows across finance,
supply chain, HR, sales, marketing and service.

40 The AIdea of India: 2025

How AI changes RPA

Source: a16z

A new approach to RPA As a result, businesses that adopt intelligent

automation early stand to gain a significant edge in
GenAI is enabling new approaches to automation their respective markets.
by combining LLMs with automation workflows.
While traditional RPA relied on scripted bots to
mimic human actions, often struggling with process
changes and requiring costly implementations, Implications for companies
GenAI enables adaptable agents that can interpret
end goals and execute tasks across various data Move to an AI-first app strategy
inputs and evolving business processes. For – adaptive and interactive
example, GenAI can extract relevant information interfaces, self-improving
from emails or PDFs and input it into enterprise models and architecture that is
systems, adapting to variations in content and scalable, flexible, and can heavy
computational demands
format without needing manual intervention.
This shift has profound implications. Intelligent
automation reduces costs by eliminating the need Prepare for the coming wave of
AI embedded enterprise apps that
for constant updates and maintenance of rigid RPA
will feature cognitive services and
systems. It also unlocks new productivity gains, adaptive learning
enabling employees to focus on strategic,
value-adding tasks while automating complex,
time-consuming processes. Furthermore,
companies can use this technology to scale Leverage next-generation
operations more efficiently, enhance customer approaches to automation (RPA)
experiences, and create competitive differentiation and integration
through smarter workflows and decision-making.

The AIdea of India: 2025 41

 Advent
Moving to an AI-ready data
foundation of new
technology
Data is the lifeblood of AI initiatives, as it forms
stacks is compelling organizations to
the foundation on which models are trained,
tested and deployed. The quality, diversity and reassess their risk posture, develop new
volume of data directly influence the accuracy frameworks for managing AI in production,
and performance of AI systems. Without clean,
and address cybersecurity challenges
structured, and representative data, even the
most advanced algorithms can produce biased
or unreliable outcomes. In the context of AI, data
technologies—collectively known as the modern
enables the identification of patterns, predictions,
data stack—are simplifying data ingestion, storage,
and insights that drive innovation, decision-making
analysis, and integration, enabling businesses to
and automation.
unlock the potential of their data.
In India, with its explosion of population scale
digital platforms, the difference between winners
and losers in the AI race will in many cases boil Preparing data for AI
down to the ability to procure and leverage diverse As businesses increasingly adopt GenAI, they face
datasets that address unique challenges. a critical challenge: most data is not well-prepared
Enterprises are rapidly building the data for AI implementation. While organizations
fabric essential for enabling model-first digital often have vast amounts of structured and
transformation with over 68% of surveyed unstructured data, they lack effective systems for
Enterprises having made meaningful strides managing and structuring it. Comprehensive
towards data readiness for AI use. To move beyond metadata – data about data - is equally crucial,
commodity model performance, companies are as it provides the context necessary to deliver
focusing on identifying, nurturing, and leveraging high-quality, timely, and relevant information.
datasets that offer competitive advantages. Without these foundational systems, data quality
Simultaneously, enterprise-grade and governance issues undermine productivity and
decision-making.

Organization’s readiness in adopting

GenAI: Data (platform readiness) Establishing a robust data framework
To fully harness AI technologies, businesses must
Enterprises in India are at different stages of data
readiness, with only a few at a mature level implement comprehensive data governance
frameworks. This includes ensuring data accuracy,
3% consistency, and seamless integration across
16% departments. Our recent report, ‘Data 4.0: Making
your data AI-ready’, highlights the need for agile
data architectures and frictionless frameworks,
advocating a shift from siloed data repositories to
unified platforms capable of managing complex
data landscapes and generating actionable insights.

23%
Role of data pre-processing
Data preparation is a critical step before feeding
information into AI models. For example,
19% 39% implementing standardization rules can address
inconsistencies and fill data gaps, particularly in
sectors like healthcare. Organizations that neglect
Fully ready Mostly ready, Partially ready, requires
and mature minor gaps enhancements
data quality put themselves at a competitive
disadvantage. A master data management system
Needs significant improvements Not ready
serving as the GenAI context is vital for establishing
Source : EY India C-suite GenAI Survey 2024 a single, reliable source of truth across all
transactional data.

42 The AIdea of India: 2025

Data requirements for GenAI and traditional AI/ML differ

Aspect Traditional AI/ML GenAI

Data scope Narrow, task-specific Massive, broad, diverse

Initially minimal; may be required to ensure

Labeling Essential, expensive
alignment later

Data quality Filtering harmful, biased or

Accuracy, consistency
focus copyrighted content

Large-scale ingestion, deduplication content

Data preparation Feature engineering, structured formats
moderation

Standard (extract, transform, and load) Distributed storage, vector databases,

Infrastructure
ETL process, feature stores retrieval-augmented set-ups

Governance and Known frameworks, Complex filtering, legal/IP checks, preventing

compliance simple anonymization sensitive data leaks

Heightened risks due to scale and diversity

Bias and ethics Manageable with balanced sampling
of training data

Evaluation Quantitative metrics on labeled sets Qualitative, human-in-the-loop, alignment tests

Lifecycle and Continuous refresh, dynamic

Periodic updates, stable pipelines
maintenance retrieval augmentation

Established ML ops, data Prompt engineers, content moderators,

Tooling and skills
engineers, labelers large-scale data curators

Unlocking AI’s full potential Implications for companies

Ultimately, effective AI utilization begins with
robust data management practices. Proactive Focus on and nurture datasets
that provide competitive
governance, quality assurance, and seamless
advantage- these will yield
data integration are essential for navigating the better-than-benchmark model
complexities of data readiness for AI. performance

AI
Implement a robust data
governance framework - ensure
-ready data is a competitive accuracy, security, and compliance
of the data estate aligned with
differentiator. Companies regulations like India’s Digital
that focus on unique datasets Personal Data Protection Act

and robust data governance will

unlock better-than-benchmark AI Implement a scalable data
performance Infrastructure – Invest in cloud
platforms or data lakes, real-time
processing tools, and AI-ready
hardware like GPUs/TPUs to
handle large datasets

The AIdea of India: 2025 43

 Increase
Getting your people ready for AI
in
Change management is a critical component to deepfake
achieving the most out of GenAI. EY estimates
that GenAI could impact 38 million jobs in India. scams and AI-driven cybersecurity
We discuss our assumptions, methodology and threats highlight the urgent need for
implications in the chapter on AI and jobs. real-time governance, explainable AI,
Our enterprise survey suggests that only 3% of and proactive cyber defenses
Indian enterprises possess the in-house talent and
resources to make the most of what AI has to offer.
Enterprises need to address the gap through a mix Finally, the productivity gains from GenAI will be
of programs to hire and retain high quality AI talent unequally distributed with certain functions such
or by seeking help from third party experts in the as Call Center Management seeing significantly
short term to bridge the gap. greater gains than R&D. In certain instances, AI
At the same time, reports suggest that over will replace human effort while elsewhere it will
92% of Indian professionals are leveraging AI at augment or amplify the same.
work, surpassing the global average of 75%. This Enterprises should systematically evaluate the
suggests that educating the rank and file on how role of AI at the level of functions, jobs and tasks
to use GenAI well and where the risks lie, along in order to re-orient and re-skill their workforce to
with arming them with GenAI tools appropriate get the most out of what GenAI has to offer. This
for citizen developers, may be one approach to is discussed further in our chapter ‘Transforming
encouraging and scaling GenAI adoption. work with GenAI’.

Organization’s readiness in adopting Implications for companies

GenAI: Talent (resource availability)
AI expertise is a key need for most enterprises as they
undergo GenAI transformation Prepare your organization for
GenAI by educating them across
16% rank and file
3%

Scope out and prepare to hire GenAI

specifi- talent or expertise to deliver
your transformation initiatives
22%

Systematically explore what GenAI

can deliver for your organization
24% 35% through a bottom up evaluation
of each job and task rather than
Extensive expertise and Talent exists but a top-down brainstorming for
resources for effective insufficient to support GenAI ideas and use-cases
deployment all initiatives
Have some skills but need Aware of skills but lack
significant investment them and have no
acquisition strategy
Have not thought about specific
GenAI skill requirement

Source : EY India C-suite GenAI Survey 2024

44 The AIdea of India: 2025

Confronting the changing
frontier of risk Ensuring accuracy
in AI systems
As GenAI adoption becomes more prevalent, it
Ensuring the accuracy of GenAI systems
requires a nuanced and well thought out approach
is crucial in both consumer and enterprise
to understanding and tacking the underlying
contexts. Intelligent assistants like Amazon
risks. Traditional risk management frameworks,
Alexa, Microsoft Bing Chat, Google
designed for deterministic systems, often fall short
Assistant and Apple Siri, for example, have
in addressing the dynamic and probabilistic nature
long faced criticism for inaccuracies. In
of AI models. GenAI systems can inadvertently
enterprise settings, such errors can prove to
introduce biases, especially in sensitive applications
be much more expensive both financially as
like credit scoring, potentially leading to the
well as leading to injury or loss of life in an
exclusion of underserved groups.
industrial setting.
To mitigate such risks, organizations must
In 2022, Air Canada’s AI-powered customer
implement real-time governance mechanisms,
service chatbot provided incorrect
including anomaly detection systems, automated
information to a customer seeking a
compliance checks aligned with regulations, and
bereavement fare. The chatbot erroneously
the development of explainable AI models to ensure
stated that the customer could apply for
transparency and accountability. Similarly, the
a refund within 90 days after booking,
European Union’s AI Act, as an example, aims to
contradicting the airline’s actual policy,
regulate AI systems based on their risk levels, from
which does not allow refunds for travel
minimal to high-risk applications. Addressing issues
already completed. When the customer
like AI hallucinations through improved techniques
applied for the refund, it was denied,
and better training data filtering enhances the
leading to a legal dispute. The Canadian
reliability of AI systems in critical tasks such as
Civil Resolution Tribunal ruled against Air
medical diagnostics and financial predictions.
Canada, ordering the airline to honor the
Implementing Responsible AI practices is vital for
refund and emphasizing that companies are
regulatory compliance and maintaining trust with
responsible for the information provided by
consumers and stakeholders. These topics are
their AI systems
discussed in greater detail in our chapter on Policy
and Responsible AI.

The AIdea of India: 2025 45

46 The AIdea of India: 2025
 Implications for companies
In January 2024, a Hong Kong-based
employee of the UK engineering firm Arup New risk management frameworks
fell victim to a sophisticated cyber scam are needed to address the
involving deepfake technology, resulting dynamic and probabilistic nature
in a loss of approximately US$25 million. of AI models
The fraudsters used AI to create
hyper-realistic video and audio deepfakes,
impersonating senior company executives Real-time governance mechanisms
during a video conference. Convinced such as anomaly detection systems
of the representations’ authenticity, the and explainable AI models ensure
employee proceeded to transfer funds transparency and accountability
across multiple bank accounts as instructed.
The deception was uncovered when the
employee later verified the transaction
with the actual executives. Arup promptly Better techniques and training
reported the incident to Hong Kong police, data filtering add to the reliability
who have classified the case as “obtaining of AI systems
property by deception” and are actively
investigating it. This event underscores
the escalating threat posed by deepfake
AI-powered security solutions,
technology in cybercrime, highlighting the
cyber defense strategies and
urgent need for organizations to enhance continuous learning across the
their cybersecurity measures and employee organization can minimize risk
training to detect and prevent such
sophisticated fraud attempts.

GenAI’s ability to produce highly realistic Traditional cybersecurity training and measures,
synthetic content introduces significant challenges while essential, may not suffice against the evolving
in distinguishing genuine information from landscape of AI-driven threats. Cybercriminals are
fabrications. This capability can be exploited to increasingly leveraging AI to execute sophisticated
spread false narratives, manipulate public attacks that can bypass conventional defenses,
opinion, and erode trust in institutions. rendering static security protocols inadequate. To
Additionally, GenAI can be leveraged to craft address these challenges, organizations will need
sophisticated phishing attacks, generate to adopt AI-powered security solutions capable of
malicious code, and create deepfakes, amplifying real-time threat detection and response, implement
the capabilities of cybercriminals and posing proactive cyber defense strategies, and foster
substantial security risks. continuous learning environments to keep pace
with the rapidly changing threat landscape.

The AIdea of India: 2025 47

Chapter 3

Transforming work
with GenAI:
EY India jobs study

48 The AIdea of India: 2025

The AIdea of India: 2025 49
Chapter 3

Transforming work with GenAI:

EY India jobs study

Unlocking the productivity

An EY study on analysing over revolution
10,000 tasks across critical industries
that contribute to the Indian economy AI, particularly GenAI, is reshaping the world of
work, heralding a transformative era for the Indian
reveals that 24% of tasks can be economy. It is set to transform every job, from
fully automated, while time spent farm labor to the role of a CEO, unlocking immense
on another 42% can be significantly potential for productivity and economic gains.
India’s workforce, therefore, is poised for significant
reduced, freeing up 8-10 hours per transformation, with every job in the organized
week for corporate workers sector expected to experience a paradigm-shifting
impact over the next 3-5 years.
This translates to a productivity boost To realize this potential, organizations must act
decisively, moving beyond piloting use cases,
of 2.61% by 2030 in the organized
to redefine processes, job roles, and KPIs and
sector affecting 38 million Indian skilling journeys. The government needs to foster
employees and additional 2.82% in the an enabling environment for AI adoption while
addressing the need for workforce adaptation and
unorganized sector
skill development.
An EY study to understand the impact of AI on
The largest productivity gains from jobs aims to address critical questions in this
GenAI are expected in the services transformative era: How will GenAI redefine jobs
sector due to its higher labor share in in India? What are the key shifts across industries?
What steps must organizations in India take to
gross output, while manufacturing and achieve sustainable productivity gains? And how
construction will see smaller impacts can the Indian government facilitate a smooth
transition into an AI-augmented work environment?
By exploring these questions, the study offers a
Realizing this potential requires
strategic roadmap for India to excel in a time of
reimagining processes, redefining rapid technological progress and evolving
workflows & KPIs and reskilling the economic landscapes.
workforce. Large-scale upskilling
initiatives, supported by public-private
partnerships and AI-focused training
programs, are crucial to bridging the
skill gap

50 The AIdea of India: 2025

Productivity survey highlights: ■ Exposure (extent to which any task lends itself
to improvement in productivity through the use
of automation and GenAI)
■ Complementarity (degree to which the
AI-driven transformation is expected to task will require human oversight and
impact 38 million employees across India’s therefore GenAI augmented)
organized sector ■ Intensity (frequency of tasks analyzed in
granular time units to estimate volume
and effort)
The task analysis has been aggregated at a
business process and industry level to generate
IT, banking, and insurance sectors are expected actionable insights. This study also analyzes the
to see significant impact, with business processes impact of AI on labor productivity using RBI’s
like customer service and software development KLEMS database, modelling the impact of labour
experiencing fundamental shifts productivity improvement on gross output. For
more details on the methodological framework,
please refer to the annexure.

This could potentially drive a 5.44% boost in Understanding productivity gain

productivity for the Indian economy at the task level

The analysis of tasks across exposure,

complementarity and intensity generates an
actionable framework for unlocking a ‘productivity
uplift’ at a job role, business process and
Research approach and organization level. Based on this analysis, the
methodology impact on tasks can be classified into three
categories:
For the purpose of this research study, we have Automate (tasks, if not already done so, can
defined the following terms: be completely automated or eliminated from
■ ‘Tasks’ are discrete individual activities the workflow)
performed using human skills and tools to Augment (tasks can be enabled by GenAI,
achieve desired outcomes leading to significant reduction in time to
■ ‘Jobs’ are structured collections of these tasks, perform them)
encompassing a set of responsibilities and Amplify (tasks that can be transformed,
duties to achieve organizational objectives requiring elevated outcomes, cognitive
■ ‘Business processes’ are sequences of processing and/or interactions)
interlinked tasks and activities designed to Our analysis across multiple sectors reveals that
produce a specific outcome that drives broader approximately 65% of tasks across industries have
organizational goals the potential for automation or augmentation.
As the foundation of the study, we have studied The greatest potential for transformation exists in
over 10,000 tasks across critical industries that sectors like IT, banking, insurance, and telecom,
contribute to the Indian economy. Every task is where AI is set to reshape operational efficiency
analyzed on these factors: and drive value creation.

The AIdea of India: 2025 51

Productivity gain insights
This exhibit illustrates the exposure and complementarity to AI on the x-axis and the intensity of doing
the same task on the y-axis. It highlights areas with the highest potential for return on investment and
opportunities for redefining job roles.

Amplify
(Tasks that can be transformed,
requiring elevated outcomes,
cognitive processing and/or
interactions)
A few examples are:
■ Investment decisions
■ Strategic business planning
■ Devising annual operating plans
■ Product design
■ Pricing strategy
■ Product diversification strategy
■ Acquisition or merger decisions
■ Personalized patient care
■ Counselling
■ Delivery services
■ Home technical services

Automate Augment
(Tasks that, if not already done so, can (Tasks that can be enabled by GenAI, leading to
be completely automated or eliminated significant reduction in time to perform them)
from the workflow)
A few examples are:
A few examples are: ■ Competitor analysis
■ L1 customer service ■ Loan structuring
■ Scheduling meetings or reminders ■ Market forecasting
■ Invoice reconciliation ■ Fraud detection
■ Resume screening and prioritization ■ Supply chain optimization
■ Order tracking ■ Event planning
■ Website analytics ■ Social media strategy
■ Basic IT support

Automate
Tasks like L1 customer service, scheduling appointments and reminders, matching and reconciling
documents and invoices, content creation etc. are either fully automated or on track to be phased out.
These account for 24% of repetitive, transactional activities.

Augment
Tasks such as competitor analysis, loan structuring, forecasting, fraud monitoring, event planning etc.
can be significantly accelerated with GenAI. These comprise 42% of tasks, which necessitate a strong
human-AI collaboration to deliver superior outcomes in lesser time.

Amplify
Tasks such as personalized patient care, counselling, hospitality services, delivery services, home
technical support as well as strategy and business decision making will increasingly require elevation
through higher order thinking, personalization, and empathy.

52 The AIdea of India: 2025

Understanding productivity gain Our estimates indicate that nearly 38 million
employees will be impacted. The integration of AI
at the business process level into our work environment has the potential to
free up 8-10 hours per week for every corporate
The pace of workforce transformation is likely to worker, over the next 3-5 years. This freed-up
accelerate, given increases in the potential for capacity presents a transformative opportunity for
technical automation and GenAI across major organizations and industries to reinvest in growth
business processes. and innovation. Businesses can leverage this
productivity boost to explore new markets,
expand into untapped categories, enhance
Understanding productivity gain customer experience, or streamline operations
to reduce costs.
at the industry level
At the broader economic level, India can open up
Our analysis based on the impact of AI adoption
new pathways for innovation, strengthen its global
on labor productivity improvement by the
competitiveness, and develop solutions customized
organized sector suggests potential to increase the
to its distinctive socio-economic context.
productivity of the economy by 2.61% up to 2030.
Further productivity improvement of up to 2.82% is The largest gains are projected to accrue in the
possible with the unorganized sector adopting AI, services sector, as the share of labor in gross
resulting in combined potential productivity output tends to be higher in this sector. The impact
gains of 5.44%. on manufacturing and construction output will be
lower, as the labor income share in gross output

What GenAI adoption means for business processes

These charts summarize the productivity uplift potential across several business processes ranging from call
center management where 80% productivity enhancement is expected to new product development where
productivity uplift could be muted, and to the tune of 17% tends to be lower.

80% 61% 45% 44% 41%

Call center Software Content development Customer service Sales and
management development and distribution marketing

41% 38% 36% 25% 17%

Supply chain band Store operations Production and Quality and regulatory New product
merchandising maintenance compliance development

The AIdea of India: 2025 53

However, even in sectors with a low income share While some organizations may still struggle with
of labour in gross output, additional productivity unclear AI strategy, evolving AI regulation, and
gains could accrue as AI applications enable talent inadequacy, investments in AI are on the
more efficient capital deployment and resource rise. It is seen as one of the most revolutionary
utilization, such as energy and materials. technological advancements in recent history and
as a multi-year opportunity. AI impact is just about
starting to show up in corporate bottom lines, with
Realizing productivity a small percentage of companies already reaping
the rewards of AI and GenAI.
gains: What should
In India, organizations have progressed
organizations do? from AI helping answer questions, to AI
making predictions, to GenAI agents. While
In the past, technology advances and automation the ability to intelligently automate tasks and
largely impacted manual labor and process-centric gain instant insights has the potential to boost
knowledge work. This time around, it is different. workplace productivity, translating this potential
The script has been flipped; GenAI will change what to profit is still a discovery. Some companies may
is generally called knowledge work — everyone from experience underwhelming results even after a
entry-level number crunchers to seasoned heads of year of investment in such initiatives, leading to
business units and C-suite executives will see their disillusionment or a loss of momentum in their
job evolve over the next decade. transformation efforts. But this is not the time to
let go, stand on the sideline, or wait and watch.

Productivity gains across key sectors

This graph illustrates the labor cost as a percentage of gross output on the x-axis and the percentage
productivity improvement through AI on the y-axis. The size of the bubble represents the potential labor
efficiencies created by AI for the industry.

50%
productivity significantly

45%

IT | 19%
40%
AI enhances

Retail | 5%
Productivity potential

Banking | 9%
35%
Pharma | 2%
Insurance | 8%
30%
Telecom | 5%
Automobile | 2%
productivity marginally

25% Metals and Mining | 4%

Healthcare | 13%
20%
Media and Entertainment | 5%
AI enhances

15%

10%
0% 10% 20% 30% 40% 50% 60%

Labor cost by gross output

Labor plays Labor plays a
a smaller role larger role

54 The AIdea of India: 2025

Introducing a structured approach to unlock productivity potential
By embracing the following tenets, organizations can take the necessary steps to bolster confidence, build
trust, and open the door to a new age of productivity

Design thinking to reimagine processes focusing on:

■ Increased efficiency and productivity
■ Better data-enabled decision-making
■ Cost reduction
GenAI strategy ■ Enhanced accuracy
■ Identifying GenAI use cases ■ Improved user or customer experience
■ Feasibility study
■ Portfolio management for Reimagine business processes Redefine tasks
GenAI investments
■ Build or buy technology Reconstruct processes basis
decisions tasks analysis:
■ Automate (tasks, if not
already done so, that can
be completely automated or
Use case: Build out and pilot eliminated from workflow)
■ Training: Building the Productivity ■ Augment (tasks that can be
realization enabled by GenAI, leading
foundation
to significant reduction in
■ Tuning: Customizing for Reinstate time to perform them)
specific tasks your vision ■ Amplify (tasks that can be
■ Testing and launch Possibilities transformed, requiring
■ Ethics considerations with GenAI elevated outcomes, cognitive
processing and/or interactions)

Potential value realization

■ Assess ROI across Reskill for new roles Repurpose work and KPIs
■ Revenue growth
■ Reskilling journeys basis the new Rewrite job descriptions with
■ Reduced operational
ask of the job (not just generic AI human-AI partnership, covering:
cost
literacy). ■ KPIs
■ Enhancing productivity
■ Foundationally, embrace holistic ■ Knowledge (technology,
■ Deeper customer
upskilling and career concepts: functional, cyber and risk,
engagement
non-linear paths, flexibility, lateral governance and ethics)
■ Decision to launch: Deploy
moves, diverse titles, function ■ Skills (AI literacy, critical
AI at scale
agnostic, personal passion thinking, emotional
enablement intelligence, empathy)
■ Experience (innovation,
learnability, ideation,
entrepreneurship)

Reinstate your vision: Unlocking GenAI’s potential

01 GenAI strategy: Identifying where GenAI

co-pilots can create real competitive
advantages is essential. A value-based
03 Potential value realization: This critical
stage evaluates whether AI investments
are delivering tangible outcomes such as
portfolio approach ensures priority use revenue growth, cost savings, improved
cases are both innovative and financially efficiency, and enhanced employee
sustainable. engagement. Promising results within
1-3 years justify scaling AI implementations.

02 Use case - Build and pilot: Develop

use cases to enhance productivity,
At this juncture, organizations can decide
whether to scale AI adoption, setting the stage for
automate tasks, and modernize customer
transformative productivity gains.
experiences. Each use case follows four
key phases: training, tuning, testing, and
ensuring ethical deployment.

The AIdea of India: 2025 55

56 The AIdea of India: 2025
 GenAI
Productivity realization

01 Reimagine business processes: Rethink

how work is performed by examining
existing tasks, identifying inefficiencies,
will change
‘knowledge
and leveraging new technologies to drive work’ — everyone from entry-level number
innovation and effectiveness. Shifting from crunchers to heads of business units and
traditional methods to transformative
C-suite executives will see their
approaches is key
jobs evolve over the next decade
02 Redefine tasks: Analyze and optimize
task construct using the 3A framework:
Automate, Augment, Amplify
Establishing AI-focused training centers
03 Repurpose work and KPIs: Elevate
critical thinking and human-centered and incentivizing industry-led upskilling
will enhance employability. For example,
tasks, redefining job descriptions and key
performance indicators to align with integrating AI into the 14,000-plus ITIs and
GenAI capabilities more than 1,100 universities can build a
future-ready workforce.
04 Reskill for new roles: Focus reskilling
efforts on preparing employees for new AI-focused Apprenticeship Scheme
job constructs, moving beyond basic AI (AIAS): Developing an ‘AI-mindset’ within
literacy. As Generations Z and Alpha enter the industrial ecosystem requires a
the workforce, redefine jobs and career large-scale AI-focused apprenticeship
paths to reflect evolving aspirations and scheme (AIAS) to skill the domestic
technological landscapes workforce. Designed in partnership with
GenAI will influence companies across all industries, the corporate sector, AIAS can build
but true success — marked by increased industry-aligned capabilities in youth.
profitability — will belong to those who prioritize Learnings from the recently announced
process optimization, workforce engagement, and Prime Minister’s Internship Scheme (PMIS)
cultural alignment alongside technology adoption. can provide a valuable template for its
successful implementation.

Leveraging Public-Private Partnerships

Upskilling at scale: Preparing (PPP): India can look to replicate the
38 million people for tomorrow’s success of PPPs in the utilization of
Digital Public Infrastructure, which helped
economy turbocharge India’s digitization journey.
PPPs can be leveraged to develop the
The government can play a significant role in ecosystem for the effective diffusion of
fostering the AI ecosystem in India ensuring AI within the economy, including skilling,
and even enhancing the competitiveness of datasets, security, compute capacity, trust,
domestic businesses. To ensure the success of and application development.
these initiatives, policy makers will have to closely
collaborate with industry bodies, corporates and As India stands at the cusp of a transformative
their leadership. era, the convergence of GenAI and human
ingenuity offers unparalleled opportunities for
Several critical areas where government productivity and growth. By embracing this change
intervention is expected to be pivotal are with a structured, holistic approach—redefining
outlined below: processes, reskilling talent, and fostering a culture
Bridging the skill gap: According to of innovation — the nation can unlock significant
the Ministry of Skill Development and economic value.
Entrepreneurship, only 51.3% of college Through a shared collaborative vision between,
final-year students in India are and decisive action from, both its policy makers
employable, highlighting a significant and corporate leaders, India can lead the way as a
education-employability skill gap. global leader in AI-powered growth
and innovation.

The AIdea of India: 2025 57

Chapter 4

Industries in
transformation
58 The AIdea of India: 2025
The AIdea of India: 2025 59
 i
Transforming transactions with GenAI:
Financial services

T
GenAI is revolutionizing financial he financial services sector saw strong
services across customer real-world adoption of GenAI and a business
case-based implementation drive in 2024
engagement, operations and compared to innovation projects in 2023. Large
risk assessments banks have been cautious in adoption of GenAI
given their large size and compliance requirements
Indian firms are adopting GenAI compared to NBFCs and a few mid-sized banks. The
majorly for voice bots, email strongest adoption has been among NBFC
automation, business intelligence, lenders and insurers. NBFCs have implemented
and workflow automation GenAI-driven solutions that have started proving
business value. Leading companies are publicly
GenAI is expected to boost stating their concrete and numerical benefits in
productivity (34% to 38% by media as well as investor and analyst presentations,
especially during the last few months of 2024.
2030 in some segments), with
significant improvements in key

Financial
areas like customer service
and operations
services
enterprises are experimenting with
various GenAI use cases to build
advanced architectural solutions that,
connect GenAI with core financial
systems like CRM, loan origination
and card management

60 The AIdea of India: 2025

Companies have also reported decrease in cost of ■ Customer calling and text driven interactions
per unit normal business operations activities from (across WhatsApp, Email, SMS, mobile app,
1/4th to 1/10th of the traditional manual operating website and voice channels),
costs using AI for applications like:

Question Question
What has been your experience with What is your organization’s overall
GenAI POC engagements? inclination to invest in GenAI?
74% respondents have initiated POCs; with 11% running 26% 16%
in production

42%
are exploring this
actively with
budgets allocated
or already
invested

32% 26%

No response for “No Plans to Invest”

Source : EY India C-suite GenAI Survey 2024
Source : EY India C-suite GenAI Survey 2024

The AIdea of India: 2025 61

Question have taken up large-scale implementation projects
Across organizations, select functions/ specifically focused on GenAI, towards the last few
departments being prioritized for GenAI months of 2024. These large scale and ambitious
projects talk of: using GenAI in all areas ranging
implementation (Mark all that apply)
from cyber security copilot and corporate lending
underwriting copilots to AI driven customer care
(inbound and outbound multi-channel) platform
implementations and on-premise GPU cloud
deployments.
Mid-sized banks that are more innovative have also
developed large scale GenAI driven orchestration
layers enabling GenAI systems to communicate
with core banking systems using Agentic AI layers
showing a longer-term strategic intent behind use
of GenAI across multiple functions.
In December 2024, IDFC FIRST Bank announced
the launch of AI-powered holographic digital
Source : EY India C-suite GenAI Survey 2024 avatar of legendary actor Amitabh Bachchan, for
customer engagement. The Holographic Extended
Reality (HXR) device has touch capability, allowing
users to interact directly with the digital avatar.
■ Credit Underwriting Summary Automation, Through this, customers can engage with the
■ Marketing collateral generation for SEO driven digital avatar and get information on bank products
website content writing, and services, including zero fee banking, monthly
interest credits, mobile banking, and their
■ Learning and development video creation, etc. current accounts.

GenAI is being used to increase the span of The bank also has developed an AI-based bank
coverage for costly manual routine activities like statement analyzer that gives deeper level insights
customer care call quality evaluation, percentage to customers on their spending habits, inflows, and
of audit evaluation samples coverage, compliance transactions.
queries and checks, etc.
Additionally, NBFCs are aggressively building Question
conversational business intelligence bots that
Where has AI been impactful?
understand the “vocabulary of the orgs” to query
(Mark all that apply)
business operations data to go beyond static
dashboards and ask n-th level deep dive and AI has been most instrumental in driving customer
dissection questions for operations productivity satisfaction and reducing costs
and profitability enhancements as well as
diagnostics of issues to focus upon.
Tata AIA Life, for example, has launched
Tasha bot that has completed a sizable number of
7.5 million customer interactions with a remarkable
goal completion rate of 98%. The bot allows
policyholders to access over 60 services such
as policy information review, payment updates,
document downloads, and profile updates digitally
using AI. Additionally, it also acts as a sales coach
to the frontline salespeople helping them with lead
management and prioritization, deliver tailored
advice and provide training and performance
information.
While banks have been relatively conservative in
number of production-level use cases, big banks Source : EY India C-suite GenAI Survey 2024

62 The AIdea of India: 2025

Overall, there was concrete on-ground progress with Email automation:
proven and significant business benefits in 2024, GenAI agents are automating email
delivered by GenAI systems at production scale management by classifying queries,
with large-scale Indian NBFCs at the forefront of allocating tickets and drafting initial
innovation, followed by midsized innovative banks. responses for human review. For example,
We expect large scale banks to also trace a similar Tata Capital has successfully implemented
trajectory in 2025. this to achieve a 20% cost saving while
reducing resolution time from one day to
20 minutes.
GenAI use cases Conversational business Intelligence:
Business leaders can now query real-time
In the financial services sector, GenAI driven metrics — such as branch performance or
innovation can be broadly categorized by key NPA rates — through conversational GenAI
adoption areas in different categories: tools, which provide data insights using
Voice agents for customer interaction: visual formats like bar charts and heat
Over the past year, many NBFCs have been maps. Custom-built solutions
deploying GenAI-powered voice bots to enable advanced analyses such as
enhance customer engagement: correlation studies, simulations and
root-cause diagnostics.
■ “Call-me” button support: Customers
can request a callback to learn more Complex workflow automation:
about products or policies. GenAI automates tasks traditionally
handled by experienced employees, such as
■ Soft-bucket collections (0–30 days insurance claims processing, underwriting
past due): AI bots engage customers in and credit memo creation. These AI
free-flowing conversations to address agents can analyze PDF reports, financial
payment challenges and encourage statements and scanned documents to
timely EMI payments. extract critical data points efficiently.
■ Recovering incomplete digital journeys: Enterprise copilots:
AI agents can reconnect with customers GenAI copilots are gaining traction as
who abandon applications mid-process productivity tools, providing employees
to resolve issues and drive completions secure access to LLMs for tasks
within 30 seconds of drop-off. like referring to internal policies,
FAQs and circulars.

The AIdea of India: 2025 63

Case Study: Bajaj Finance uses future. Bajaj Finance is aspiring to increase its
direct sales conversion rate as well as cross-sell
GenAI for transforming customer
conversion rate by 3X using conversational sales
care, sales and onboarding copilot based in GenAI.

Bajaj Finance has unveiled a 5-year all-round AI

adoption roadmap. This year, it has attributed a
GenAI’s impact on productivity
saving of INR150 crore to AI compared to last year
in multiple business operations areas. It is using India’s financial services sector, employing millions
across banking, insurance, and asset management,
AI-driven bots to provide hyper personalized loan is set for a significant productivity transformation
offers and cross-sell offers. The company driven by GenAI.
aims to grow its targeted messaging base to
Our study analyzing over 700 roles in the sector
500 million from 160 million at present. (more than 300 in banking and 400 in insurance)
Additionally, it plans to use AI to allow customers projects that GenAI will drive a 34% to 38%
to evaluate which loan product is most suitable for productivity improvement, paving the way for
them, book an appointment with a sales manager growth and operational transformation in the
and even automate the eKYC procedure in the industry.

GenAI impact on productivity across banking and insurance value chain

Size of bubble represents the relative headcount in the business process for the sector

Source: EY India jobs study

64 The AIdea of India: 2025

Mid-sized
are developing orchestration layers with
banks
Challenges and way ahead
Despite its transformative potential, the
financial services sector faces challenges
Agentic AI, enabling seamless integration in scaling GenAI, primarily due to strict
with core banking systems and regulatory frameworks and data privacy
concerns, such as:
showcasing a strategic commitment to
leveraging AI across multiple functions for Data localization: Financial services
long-term growth firms must ensure LLM endpoints
are hosted within India, through
on-premise solutions or India-hosted
hyperscaler endpoints
Banks Data privacy: Firms must avoid
sending Personally Identifiable
Functions with highest potential for
Information (PII) to GenAI APIs
productivity impact include banking
by using PII redaction tools and
operations (44% to 46%), sales and
anonymized data
customer service (38% to 40%) and credit
and collections (34% to 36%), which Cybersecurity: Hosting LLM
can together account for 40% to 42% of architecture within Virtual
productivity improvement Private Cloud (VPC) environments
is essential for securing
Functions like channel management,
enterprise systems
products and services portfolio are also
expected to see significant productivity Indian financial services firms are
improvement accelerating innovation in GenAI with a focus
on hyper-personalized financial products
Insurance (for example, personalized credit cards) and
GenAI-driven customer interactions.
Customer service (~48%), new business By embracing conversational AI and
(46%), and sales and partner (45%) intelligent banking solutions while
management can account for 45% to 48% of ensuring compliance, this sector is poised
overall productivity improvement to lead the global adoption of GenAI
Investments and new product development, technologies, driving a seamless and
along with claims management, are customer-centric future.
also key areas offering opportunity for
productivity improvement

The AIdea of India: 2025 65

ii
Redefining consumer
engagement with GenAI:
Retail, consumer and
e-commerce

G
GenAI is transforming retail, enAI has become a mainstream technology
consumer, and e-commerce across the consumer, retail and e-commerce
sectors. The EY Future Consumer Index,
through innovation in product released in July 2024, highlighted growing
development, sales and marketing acceptance of AI-driven solutions, particularly
and customer engagement in the Asia-Pacific region. Despite initial
apprehensions regarding risks such as bias and
GenAI is expected to boost reputational damage, companies are adopting
productivity by 35% to 37% by robust risk management frameworks to scale GenAI
2030, with significant impact enterprise wide.
on insights-driven pricing and Global players are leveraging GenAI to transform
promotions and customer business processes and consumer experiences.
experience Companies like H&M are enhancing sustainability
initiatives, product development, and marketing;
Future GenAI success will Coca-Cola is optimizing delivery routes and stock
depend on balancing automation management; and Amazon Go is enabling
with human experiences and

GenAI
overcoming skill gaps

is set to make the

retail sector more
agile, efficient and customer-centric by
connecting diverse processes including
procurement, supply chain and quality
management

66 The AIdea of India: 2025

cashier-less checkout. Similarly, Zalando employs In India, companies are leveraging GenAI-driven
fraud detection systems, Macy’s uses dynamic solutions to address diverse consumer needs,
pricing algorithms, and Sephora offers AI-powered drive efficiency and innovate, enabling growth and
personalized skincare consultations. These adaptability even with resource constraints.
applications showcase the transformative
potential of GenAI to reshape operations and
consumer interactions.

Question Question
What has been your experience with What is your organization’s overall
GenAI POC engagements? inclination to invest in GenAI?
48% respondents have initiated POCs; with 16% running
in production

Source : EY India C-suite GenAI Survey 2024 Source : EY India C-suite GenAI Survey 2024

The AIdea of India: 2025 67

Question Merchandizing: Designers are using GenAI
Across organizations, select functions / to create store layouts, mood boards and
departments being prioritized for GenAI visual merchandising designs, optimizing
time and improving efficiency.
implementation (Mark all that apply)
Pricing strategies: GenAI is enabling
real-time competitive pricing analysis,
helping brands respond to market shifts
and maximize revenue while staying aligned
with consumer expectations.

Customer service: GenAI-powered

chatbots and voice assistants are delivering
24/7 customer support, enhancing
interactions by understanding context
and regional languages. For instance, the
Indian captive of a global consumer goods
company is optimizing customer care
Source : EY India C-suite GenAI Survey 2024 through a GenAI platform that provides
guided workflows, pre-emptive alerts and
automated call summaries, improving
resolution rates and reducing
GenAI use cases handling times.

Security and anti-counterfeiting: A

GenAI is transforming the consumer, retail, and global consumer goods company in India
e-commerce industries across a wide range is using GenAI-powered tools to compare
of use cases: product data against verified databases,
Product development: GenAI is driving issuing high-alert notifications to combat
innovations by continuously analyzing counterfeit goods and protect
consumer preferences and behavior, brand integrity.
enabling the creation of more relevant and
Manufacturing: GenAI is improving in-line
appealing products.
quality control, enabling digital twins for
Marketing: GenAI is empowering predictive maintenance and service, and
hyper-personalized campaigns by enhancing worker training programs.
targeting micro-segments, creating precise
Supply chain and inventory management:
audience profiles and delivering tailored
GenAI is optimizing demand forecasting,
content through segment creation and
route planning and warehouse operations,
content adaptation.
reducing errors and increasing
Sales and recommendations: AI overall efficiency.
recommendation engines are enhancing
Post-sales: Enhanced feedback
shopping experiences. For example,
mechanisms and predictive maintenance is
the Indian captive of a global retailer
allowing brands to reach out to customers
is developing an AI platform to assist
through chat platforms for reminders on
consumers with home improvement needs,
upcoming service schedules.
while the Indian arm of a global snack
food company is providing its sales force Last-mile connectivity: GenAI is
with real-time insights on retailers’ buying streamlining last-mile logistics by optimizing
patterns. Platforms like Amazon’s Rufus, demand planning, route management and
Flipkart’s Flippi, and quick-commerce warehouse automation to ensure seamless
apps such as Zomato, Blinkit, and delivery experiences.
Swiggy are using GenAI for personalized
product suggestions, intuitive customer Through these and other use cases, GenAI is
interactions, and Indian language queries. continuously reshaping the industry by driving
efficiency, fostering innovation and delivering
personalized consumer experiences.

68 The AIdea of India: 2025

 Where to expect AI adoption in the future

Short-term Mid-term Long-term

■ Increased AI-powered ■ Widespread adoption of AI- ■ Complete digital
customer service chatbots driven demand forecasting transformation of the
■ Enhanced supply chain ■ Automated inventory supply chain
optimization using management and ■ Autonomous
predictive analytics replenishment manufacturing and logistics
■ Personalized marketing ■ AI-optimized ■ AI-driven product
and advertising manufacturing and customization and
■ Improved product production planning personalization
recommendation engines ■ Increased use of computer ■ Integration of augmented
■ Expanded use of AI-driven vision for quality control reality (AR) and virtual
market research and ■ Development of reality (VR) in customer
consumer insights AI-powered product design experiences
and development ■ Emergence of new
business models driven by
AI-enabled services

Case study: Aditya Birla Fashion ■ Product recommendations: AI algorithms

recommend products based on browsing
and Retail patterns and in-store preferences.

Aditya Birla Fashion and Retail (ABFRL) has

implemented GenAI solutions to accelerate product ABFRL has reduced digital print creation time from
development and enhance cataloguing for online 8 hours to 2 hours, achieved approximately
and offline channels. Key initiatives include: 12% improvement in PDP-to-cart conversions, and
cut product description generation costs by over
■ Trend and competition analytics: Automated 90%. The sample creation process, which previously
bots scrape the web for fast-moving trends and took three weeks, now takes 48 hours due to a
competitive intelligence. 3D ecosystem. Accuracy on prediction of success
■ Mood boards and design creation: Internally metric for future designs based on test data was
developed tools, powered by fine-tuned LLMs, more than 70%.
generate mood boards and product designs.

AI
■ Success rate prediction: AI models analyze
historical data to predict the success of new
designs with over 70% accuracy. -based demand forecasting,
■ AI-led cataloguing: Computer vision tools optimized manufacturing and
extract product attributes (for example, collar production planning and product design
type, print type) to auto-populate marketplace are expected to spread in the retail sector
attribute sheets.
in the medium term.
■ SEO-optimized product descriptions:
GenAI generates compelling product
descriptions that attract online shoppers and
improve search visibility.

The AIdea of India: 2025 69

Question
GenAI’s impact on productivity
Where has AI been impactful?
(Mark all that apply) The Indian consumer, retail, and e-commerce
AI has been instrumental in reducing costs,increasing sectors, which contribute significantly to the
revenue and driving customer satisfaction economy and employ millions, are undergoing
a profound transformation with the adoption of
GenAI. As companies roll out enterprise-wide AI
solutions, productivity gains are expected to be
substantial.
According to EY India, GenAI is projected to deliver
a productivity impact of 35% to 37% by 2030
across roles in the retail sectors, driving operational
transformation and growth. While GenAI is
expected to have an impact on most business
processes, a few stand out when measured by the
automation impact on tasks as share of headcount
deployed to deliver the task:

Source : EY India C-suite GenAI Survey 2024

GenAI impact on productivity across retail value chain

Size of bubble represents the relative headcount in the business process for the sector

Source: EY India jobs study

70 The AIdea of India: 2025

■ Insights-driven pricing and promotions and
customer experience are expected to have
maximum productivity impact of 40% to 45% Challenges and future
followed by store operations accounting for roadmap
38% to 40% of productivity improvement
Looking ahead, as GenAI evolves, its ability
■ Merchandising operations and marketing are to dismantle technology silos and enhance
also expected to see significant productivity collaboration across ecosystems will be
improvement key. By connecting diverse processes—such
Innovations in these processes are enabling as procurement, supply chain visibility
businesses to achieve greater cost efficiency, and quality management—GenAI is set to
operational excellence, and customer satisfaction. make the industry more agile, efficient and
customer-centric. However, maintaining a
balance between automation and
human-centric experiences will remain
essential for long-term success.

The AIdea of India: 2025 71

iii
Re-imagining India’s healthcare with GenAI:
Healthcare

G
Healthcare providers in India are lobal leaders in AI and healthcare are
leveraging GenAI for workforce driving innovative projects that highlight the
transformative power of GenAI. Healthcare
efficiency, predictive analysis, providers worldwide are embracing GenAI to
clinical decision support and improve patient care, optimize operations and
supply chain management foster innovation. From personalized treatment
plans and automated diagnostics to virtual health
GenAI is expected to boost assistants and drug discovery, GenAI is reshaping
productivity by 30% to 32% by traditional healthcare models and enhancing
2030, largely driven by outcomes across the industry.
non-clinical operations, clinical Healthcare institutions are also integrating GenAI
operations in addition to sales, into their systems. For example:
branding and community outreach ■ Healthians launched WelliO, a personal health
chatbot, to provide users with tailored health
Despite challenges like data insights and recommendations. WelliO interacts
sensitivity and legacy systems, with users to answer health-related questions
GenAI is poised to revolutionize and suggest relevant tests, making healthcare
India’s healthcare sector, driving guidance more accessible. Healthians also
rolled out an AI-powered IVR Bot, enhancing
productivity and improving
customer service by efficiently managing calls
patient outcomes and resolving issues in real time. By harnessing
AI in these areas, Healthians aims to transform
the healthcare experience and make healthcare
more accessible and user-centric.

■ Apollo 24|7 , digital platform of Apollo Hospitals

group, worked with Google Cloud to build a
Clinical Intelligence Engine (CIE) using Google
Cloud’s Vertex AI and GenAI models. This

72 The AIdea of India: 2025

 enables doctors to identify the next best action efficiency. It is being used to tackle critical
for patients during consultations. The CIE challenges like increasing diagnostic precision,
service leverages data from Apollo Hospitals and advancing telemedicine, and extending healthcare
LLMs from Google Cloud to create a proprietary access in rural areas.
solution in which all patient data is kept securely
within the hospitals’ systems. Question
What has been your experience with
■ Cleveland Clinic leveraged AI-powered GenAI POC engagements?
predictive analytics in cardiology to enhance
early detection of cardiovascular diseases. 66% of the respondents have initiated POCs
This approach improved risk assessment
accuracy, enabled earlier interventions, reduced
healthcare costs, and successfully identified 22%
patients requiring advanced care. 45%

■ Johns Hopkins Medicine implemented the

AI-powered Targeted Real-Time Early Warning
System (TREWS) to detect sepsis early by
analyzing real-time patient data within its
EHR system. This innovation improved early
detection, reduced sepsis-related mortality,
shortened hospital stays, and enhanced 33%
clinical workflows, setting a benchmark for
AI in critical care.

GenAI adoption in India’s healthcare sector is

accelerating as providers modernize their systems,
enhance diagnostics, and boost operational No response for ‘GenAI running, in production’ and ‘POCs
completed-succesful POCs lined up for production’

Source : EY India C-suite GenAI Survey 2024

The AIdea of India: 2025 73

74 The AIdea of India: 2025
Question for early disease detection and enabling
What is your organization’s overall focused screening and prevention. It
inclination to invest in GenAI? can forecast pharmaceutical and supply
demand to optimize stock levels, minimize
22% 22% waste, and detect supply chain issues,
ensuring timely availability of essential
healthcare provisions.
Medical imaging and diagnostics

22%
are exploring this
GenAI can accelerate the analysis of
medical images and lab results, improving
accuracy in disease detection. AI-driven
actively with
budgets allocated imaging tools assist clinicians in making
or already faster, more precise diagnoses.
invested
GenAI tools are increasingly being deployed in
healthcare organizations, either at pilot or full
implementation stages, to address critical needs in
22% 34% clinical assistance and revenue and
finance processes.

01 One prominent use case is the medical

documentation co-pilot, which helps
hospitals create and manage patient
No response for ‘Bullish, investing’ records more efficiently. This automation
reduces time spent on documentation,
Source : EY India C-suite GenAI Survey 2024
allowing healthcare professionals to focus
more on patient care.

Use cases: Transforming the

healthcare value chain
02 Another application is in revenue cycle
management, where GenAI solutions
streamline administrative processes,
enhance accuracy, and lower
With most healthcare organizations in India operational costs.
focusing on bringing efficiencies into processes in
the short term, the efforts are towards creating
strong datasets through Electronic Medical Records
03 GenAI also plays a significant role in
improving patient engagement and
healthcare delivery through health bots.
(EMR) and the Hospital Information System (HIS), These chatbots facilitate consistent patient
which will create the foundation for implementation interactions and experience, automate
of various GenAI solutions: routine inquiries and support overall
Workforce efficiency operational efficiency.
GenAI can automate administrative
tasks like appointment scheduling,
medical documentation and patient data
04 In addition, finance process automation
powered by GenAI enables predictive
analytics for financial planning, minimizes
management, reducing the workload on errors, and supports better decision-
healthcare professionals. making, resulting in greater operational
Virtual assistants provide patients with efficiency and cost savings. Together,
basic health guidance and monitor vitals these tools are transforming healthcare
remotely, enhancing service delivery and operations, making them more effective
data utilization. Such global and local and resource efficient.
advancements highlight GenAI’s potential to
transform healthcare.
Disease surveillance
and preventive healthcare
GenAI can enhance predictive analysis in
healthcare by identifying high-risk groups

The AIdea of India: 2025 75

Question Question
Across organizations, select Where has AI been impactful?
functions / departments being (Mark all that apply)
prioritized for GenAI implementation AI has been instrumental in increasing revenue and
(Mark all that apply) driving customer satisfaction

Operations
56%
Clinical services and operations
44%
Finance
44%
Customer service
33%
Sales
33%
Source : EY India C-suite GenAI Survey 2024

Source : EY India C-suite GenAI Survey 2024

Case study: AI-powered disease

surveillance by Wadhwani AI GenAI’s impact on productivity

Wadhwani AI has partnered with the Government The Indian healthcare sector employs
of India to deploy an AI-powered media scanning over 8 million individuals and faces significant
solution for event-based disease surveillance. challenges, including workforce shortages,
As part of the Integrated Disease Surveillance infrastructure limitations, and high demand for
Programme (IDSP), this system automates the services. GenAI has the potential to address these
detection of disease outbreaks by monitoring over issues, driving transformation across hospital
100,000 media outlets across multiple languages. management, clinical operations, and
patient engagement.
The AI system continuously collects and analyzes GenAI in the Indian Healthcare sector will have
data from news reports and digital media to identify significant impact across hospital management
potential outbreak signals. This information is operations, enablement of doctors and nurses
integrated into the Integrated Health Information (who comprise approximately 45% of hospital
Platform (IHIP), enabling public health officials to staff) and overall patient experience by enabling
take timely action. Alerts generated by the system seamless data availability. Our study of 450-plus
are shared with Central Surveillance Units and roles indicates a potential productivity impact of
State Units, eventually reaching District and Block 30% to 32% by 2030 across the healthcare sector.
Units to facilitate on-ground responses. For business leaders, embracing GenAI is not just
This solution has significantly reduced the an option; it is essential for driving innovation and
time required to detect potential outbreaks, achieving a significant productivity improvement.

Healthcare
strengthening India’s ability to respond to public
health emergencies effectively.

organizations in India are creating

strong datasets as the foundation
of GenAI implementation to achieve
process efficiencies

76 The AIdea of India: 2025

GenAI impact on productivity across healthcare value chain
Growth gains Efficiency gains
Size of bubble represents the relative headcount in the business process for the sector
High

Non-clinical operations
Ease of implementation

Branding and
community outreach
Medium

Patient experience

Audit and compliance

Clinical services
and operations
Low

Low Medium High

Potential productivity impact of GenAI

Source: EY India jobs study
Key areas of productivity improvement include
■ Non-clinical operations (revenue cycle management, supply chain, control towers), clinical operations
(documentation support), patient experience (digital front doors like chatbots, personalized engagement)
and branding and community outreach (targeted marketing) can account for 30% to 35% of
productivity improvement.
■ Audit and compliance functions can also contribute to productivity improvement significantly.

Challenges and way ahead

Despite its transformative potential, GenAI adoption in healthcare faces several challenges. Data
sensitivity and privacy concerns make it difficult to integrate GenAI solutions while ensuring
compliance with strict regulations. Many healthcare providers rely on legacy systems like HIS and
EMR, which are not designed for GenAI applications, necessitating infrastructure modernization for
large-scale adoption. Additionally, there is a shortage of skilled professionals trained to implement
and manage these technologies, highlighting the need for investment in training and upskilling.
Regulatory and ethical considerations further complicate adoption, requiring robust governance
frameworks to ensure the responsible use of GenAI. Limited pilot adoption and scalability of GenAI
solutions also hinder widespread implementation, emphasizing the importance of expanding
successful pilots and demonstrating clear ROI.
To address these barriers, healthcare organizations must build strong datasets through systems
like EMR and HIS for seamless data integration and invest in modern infrastructure and AI talent
development. Establishing responsible AI practices and fostering collaboration with technology
providers, start-ups, and regulators will be critical to scaling innovations. As these solutions mature
and more use cases emerge, GenAI has the potential to transform India’s healthcare sector by
enhancing efficiency, improving accessibility, and delivering better patient outcomes.

The AIdea of India: 2025 77

iv
GenAI accelerates discovery:
Life sciences

Y
GenAI is transforming drug ear 2024 marked a groundbreaking moment,
discovery, manufacturing and as the Nobel Prize in Chemistry was awarded
to Demis Hassabis, John Jumper, and David
quality management in the life Baker for their pioneering work in protein science
sciences sector using AI models like AlphaFold, which accurately
predicts protein structures. This achievement
GenAI is expected to boost highlights the transformative potential of GenAI,
productivity by 32% to 34% by especially in the life sciences sector, where
2030, helping sales, supply chain pharmaceuticals and medtech are leveraging such
and production functions advancements to catalyze drug discovery, enhance
clinical trials, optimize manufacturing, strengthen
However, to fully harness GenAI, quality adherence and reimagine drug launches and
life sciences organizations need commercial models.
to address challenges like data Globally, life sciences organizations are leveraging
privacy, infrastructure gaps and GenAI across the value chain to accelerate drug
skill shortages development and boost productivity. For instance,
Moderna has collaborated with OpenAI to advance
mRNA medicines, while Janssen has partnered
with Syntegra to use high-fidelity synthetic data for
internal projects. Sanofi’s AI-powered logistics app
predicts inventory shortages, enabling proactive
action, and Fujifilm’s AI-driven ECHELON Synergy
system enhances MRI imaging through advanced
reconstruction algorithms.
An EY-led GenAI transformation program for a
global pharmaceutical company demonstrated
a potential 14% incremental growth in EBITDA.

78 The AIdea of India: 2025

This growth was primarily driven by R&D (29%), Use cases in India:
manufacturing and quality (27%), and marketing
and commercial operations (21%). Transforming value chain
Question
India’s life sciences sector is steadily integrating
What has been your experience with GenAI across its value chain. Building on past digital
GenAI POC engagements? transformation initiatives, Indian pharma leaders
recognize GenAI’s potential to process structured
50% of the respondents have initiated POCs; with 25%
running in production
and unstructured data, synthesizing insights to
generate molecular, textual, and visual content.
For example, Syngene, a leading Indian Contract
Research Organization (CRO), has developed a
proprietary AI platform Syn.AI™ and Sarchitect™
to expedite the drug discovery process. The
expanded capabilities of Syn.AI™ include
enhanced target identification and validation,
pinpointing the most promising drug targets for a
range of diseases.
AI start-ups are playing an increasingly crucial
role in drug discovery. For example, Boltzmann’s
platform designs novel drugs, generates
synthetic pathways for molecule creation, and
uses patient data to enhance clinical trials and
predict outcomes. Prescience Insilico accelerates
optimization and evaluation in silico drug
development using its SyMoG/AI platform and
No response for ‘POCs completed - successful POCs lined up for advanced virtual screening algorithms. Both
production’ and ‘POCs in progress’ start-ups leverage AI to address diseases
Source : EY India C-suite GenAI Survey 2024 effectively, including vaccine and antibody design.

The AIdea of India: 2025 79

Question
Question
What is your organization’s overall
Across organizations, select functions /
inclination to invest in GenAI?
departments being prioritized for GenAI
implementation (Mark all that apply)

Source : EY India C-suite GenAI Survey 2024

No response for ‘Bullish and already investing’ and ‘No plans to

invest’
Source : EY India C-suite GenAI Survey 2024

80 The AIdea of India: 2025

Use cases across the life sciences value chain can Question
be seen broadly in five areas: Where has AI been impactful?
(Mark all that apply)
Research & Development (R&D)
AI has been instrumental in reducing cost and driving
GenAI for pharmacovigilance
customer satisfaction
GenAI is transforming pharmacovigilance
by automating the detection, assessment,
and reporting of adverse drug events
(AEs) from unstructured data sources like
clinician notes, publications and social
media. It generates structured reports
compliant with regional regulatory
guidelines, minimizing compliance risks.
Additionally, GenAI supports post-market
surveillance, ensuring long-term drug
safety monitoring through efficient
processing of vast datasets.

Other use cases include ‘GenAI-enabled

R&D assistants’ to optimize synthetic
routes for target molecules and
‘Clinical agents’ that assist trial teams Source : EY India C-suite GenAI Survey 2024
in study selection, CRO evaluation and
regulatory submissions, ensuring efficiency Supply chain
and compliance. GenAI enables should-cost modelling for
procurement, optimizing pricing decisions
Manufacturing by analyzing data from diverse sources
GenAI is transforming manufacturing like research papers and market trends.
processes with tools like Golden Batch This approach reduces material costs and
Analytics, which identify optimal process enhances decision-making for new
parameters to produce consistent batches product launches.
with improved yield, cycle time and
quality. Additionally, Intelligent Process Commercial operations
Optimizers for contract development and GenAI enhances salesforce effectiveness
manufacturing organizations (CDMOs) by analyzing doctor profiles, disease
leverage GenAI to enhance recipe patterns, and micro-market trends
generation and minimize waste, ensuring to deliver personalized insights. This
“Right First Time” production. empowers medical representatives (MRs)
to provide more relevant recommendations
Quality management to healthcare professionals (HCPs).
GenAI improves quality adherence by Additionally, GenAI aids in launch planning
streamlining deviation management, by generating data-driven insights
Corrective and Preventive Action (CAPA) on molecules, suppliers and market
documentation, and root cause analysis. opportunities, reducing time-to-market
Tools like GenAI Chromatography Data and costs.

AI
Analyzers can reduce report drafting
time by 40% to 50%, accelerating product
release timelines. In Laboratory Information
Management Systems (LIMS), GenAI start-ups are playing an
co-pilots automate master data increasingly crucial role in
management, reducing errors and drug discovery in areas such as delivery
improving data consistency.
of personalized therapies, inventory and
supply management, quality management,
regulatory filings, and more

The AIdea of India: 2025 81

 Case study: Driving doctor Challenge: Training the LLM appropriately
on the company’s knowledge base and
engagement with GenAI managing the output with business level
prompt engineering. The company used an AI
A leading Indian pharmaceutical company has agent as a content editor, built in continuous
revamped its front-end tools with GenAI to feedback and self-training capability to remove
customize interactions and improve insights based hallucinations and generate contextual
on underused data such as visit logs, doctor insights. Responsible AI guard rails included an
feedback, prescription patterns, and local disease AI maker-checker agent for generated content.
insights. This creates an information ecosystem
for field reps, area business managers and
marketing teams.
Field sales representatives can use GenAI-built Impact of GenAI on productivity
analysis of non-personal aggregated data on
patient wellness, region-specific disease, and
prescription trends to generate insights and actions The Indian pharmaceutical sector, among the
tailored to each doctor. Only the top insights and largest globally, is set to achieve significant
actions are shared for quick understanding and productivity gains through GenAI adoption. With an
prevent information overload. Marketing and sales estimated 32% to 34% overall impact and 500-plus
leadership use a custom visualization tool to assess roles analyzed as per EY India’s study, GenAI is
trends at a location or cluster level and head office poised to revolutionize operations across the
teams can improve their marketing strategies and value chain.
budget allocations based on emerging trends.

GenAI impact on productivity across life sciences value chain

Growth gains Efficiency gains
Size of bubble represents the relative headcount in the business process for the sector
Manage customer service
High

Manage sales
Ease of implementation

Manage production
Medium

Manage supply chain

Sourcing and procurement

Equipment
management
Research, design and develop products
Low

Regulatory compliance

Low Medium High

Potential productivity impact of GenAI

Source: EY India jobs study

82 The AIdea of India: 2025

Globally
Sales (~40%), supply chain (~38%) and production
(~35%) can potentially account for 35% to 40%
and in of productivity improvement. Other areas, such
India, life as customer service, sourcing and product
development, also stand to benefit from GenAI,
sciences organizations are leveraging further enhancing overall sector efficiency.
GenAI across the value chain to accelerate
drug development and boost productivity

Challenges and way ahead

While GenAI offers immense opportunities, it also introduces challenges for life sciences
organizations, mostly in terms of:
Data privacy and security risks: The reliance on sensitive patient data raises concerns
about data breaches and compliance with stringent regulatory requirements
Model bias and ethical concerns: GenAI models must be designed to avoid biases that could
compromise drug safety or efficacy, requiring careful validation and monitoring
Infrastructure gaps: Many organizations lack the modernized infrastructure needed to scale
GenAI applications effectively
To overcome challenges in adopting GenAI, life sciences organizations need a structured approach
that includes implementing responsible AI practices to address bias, data security, and ethical
concerns. Additionally strategic investments in modern infrastructure and AI talent development,
coupled with collaborative ecosystems with technology providers, start-ups, and regulators will be
key. By adopting these strategies, the sector can unlock GenAI’s full potential, driving productivity,
innovation and better patient outcomes.

The AIdea of India: 2025 83

 v
Empowering innovation with GenAI:
Technology services

I
Technology service providers n 2023, GenAI established itself as a
are leveraging GenAI to unlock transformative force, becoming a strategic
priority for IT/ITeS companies globally. Its rapid
scalable innovation and adoption reflects the urgency felt by enterprises
accelerate growth to embrace this shift in the enterprise technology
landscape. Companies worldwide rushed to
From software development to understand and capitalize on this transformative
customer engagement, GenAI wave, leveraging technology service providers
is redefining how technology to conduct proofs of concept (POCs) focused on
services deliver value cost optimization, productivity enhancement and
efficiency gains.
With a 43 to 45% productivity Year 2024 saw enterprises transitioning from
boost across 520-plus roles by exploration to scaling successful pilot programs.
2030, GenAI is reshaping the GenAI became a key agenda item in boardrooms
future of work in technology and strategy discussions, with organizations
services investing heavily to capture the opportunities it
presents. Service providers, in turn, are poised to
Overcoming challenges, tech redefine their portfolios, unlock profitability and
create value across the next five years.
service providers are charting

For GenAI
a strategic path to maximize
GenAI’s potential

integration in the long term, tech services

firms are establishing AI offices, focusing
on architecture, regulatory compliance and
faster ROI through targeted POCs

84 The AIdea of India: 2025

Indian technology service companies have deploying GenAI-powered virtual agents to enhance
taken center stage in this movement, not only customer interactions and reduce operational
executing GenAI pilots for their clients but also costs. These developments signify a redefinition of
incorporating GenAI into their internal operations. business models, where AI is no longer a tool for
Many firms report automating up to 80% of experimentation but a core component of
software development tasks, fundamentally strategic planning.
altering traditional workflows. Similarly, BPOs are
Question Question
What has been your experience with What is your organization’s overall
GenAI POC engagements? inclination to invest in GenAI?
89% of the respondents have initiated POCs; with 33% 17% 49%
in production.

6% 78%
are exploring this
actively with
budgets allocated
or already
invested

28%

No response for ‘No plan to invest’

Source : EY India C-suite GenAI Survey 2024 Source : EY India C-suite GenAI Survey 2024

The AIdea of India: 2025 85

 Use cases

The traditional tech services sector is undergoing foundations, often requiring shifts to public cloud
a transformation as automation and packaged solutions, to fully realize AI’s potential. Tech
GenAI solutions reduce the need for outsourcing services firms are also partnering with clients
and managed support for routine tasks. GenAI has to establish AI offices, focusing on architecture,
accelerated product development cycles by up to regulatory compliance and faster ROI through
75%, streamlining software development through targeted POCs, paving the way for long-term
capabilities like business requirement generation, GenAI integration.
code conversion, and test data preparation, though
areas like product design, code reviews, and legacy The adoption of GenAI in the technology services
documentation still lag behind expectations. BPO sector spans a range of innovative use cases,
companies are leveraging GenAI for virtual agents categorized into three maturity horizons:
in customer service, handling tasks like C-SAT Horizon 1 (immediate impact), Horizon 2
surveys and lead generation more efficiently, (medium-term differentiation), and Horizon 3
while AI tools ensure accurate documentation (long-term autonomous operations). These
through real-time call summarization. However, applications are reshaping traditional
organizations need robust, scalable data workflows, offering unprecedented efficiency
and personalization.

Horizon 1 Horizon 2 Horizon 3

(Here and now) Differentiation Autonomous

(in next two years) operations only
exceptions managed
(in next 5 years)

■ Service or product support and

resolution
■ C-SAT survey Autonomous
Customer

■ Workforce ■
■ Sales orders and bookings escalation
care

optimization/
■ Lead generation workforce ■ Ad-hoc bundled
■ Agent behavior and call quality management offer creation
analysis
■ Call summarization, etc.
management care

■ Persona-led knowledge
Knowledge

search and generation

■ Knowledge ■ Self-creation and
aggregation and ■ Auto knowledge base enrichment of
summarization article generation for knowledge assets
shared service/contact
center study trends

■ Assisted application development

and testing
■ Business requirements generation
via copilots from text or voice
Development

inputs
■ Application
■ Legacy modernization - ■ Application
maintenance
Retrospective functional document modernization
generation for legacy code; code ■ Infrastructure as code
conversion and generation
■ Test data preparation
■ Design or prototype for products
■ Code review and bug fixing

86 The AIdea of India: 2025

Question Question
Across organizations, select functions / Where has AI been impactful?
departments being prioritized for GenAI (Mark all that apply)
implementation (Mark all that apply) AI has been instrumental in driving Customer
Satisfaction and Reducing Costs
Development
56%
Operations
39%
Quality & Testing
28%
Design & Prototyping
22%
Integration & Deployment
22%
Source : EY India C-suite GenAI Survey 2024

Source : EY India C-suite GenAI Survey 2024

Case study: Improving operational Impact of GenAI on

efficiency with GenAI productivity
A US-India technology company leveraged Azure India’s IT sector, employing over 5.5 million
Cloud and Azure OpenAI’s GPT-4-powered GenAI to professionals, is undergoing a transformation as
enhance its technician productivity and GenAI reshapes key functions such as software
service quality. development, IT consulting, BPO services, and
infrastructure management. Leveraging GenAI,
The solution integrates field service data, manuals
the technology services sector has the potential
and parts information into a unified platform,
to unlock substantial global economic value, with
enabling technicians to access contextual data
productivity improvements projected at 30% across
quickly. It identifies machines through serial
critical functions like application development and
numbers, links to past reports and provides insights
business process management.
into similar issues while facilitating peer-to-peer
technician support. According to EY India’s study of 520-plus roles in
this sector, GenAI could drive an estimated
The intuitive web application has reduced
43% to 45% productivity gain, significantly
resolution time and increased operational
enhancing efficiency and innovation in the sector.
efficiency by 25%, significantly boosting customer
satisfaction. This scalable solution is now deployed
across industries such as agriculture, automotive,
and HVAC.
01 Software development (~60%), BPO
services (~52%) and IT consulting (~47%)
could account for 50% to 60% of overall
productivity improvement

Enterprises 02 Quality and knowledge management,

program and project management along
with RMG are also opportunity areas for
driving productivity improvement.
are redefining business models as
AI has become a core component of
strategic planning

The AIdea of India: 2025 87

88 The AIdea of India: 2025
GenAI impact on productivity across technology services value chain

Growth gains Efficiency gains

Size of bubble represents the relative headcount in the business process for the sector
High

Software
development
Ease of implementation

Quality and knowledge

management
Talent and resource BPO and managed
management services
Medium

IT consulting
Sales and
account management

Program and project management in IT

Low

Low Medium High

Potential productivity impact of GenAI

Source: EY India jobs study

Challenges and way ahead

GenAI adoption presents significant opportunities, but technology service providers must navigate key
challenges to fully realize its potential. Internally, fragmented AI initiatives, outdated infrastructure, and
insufficient data readiness often undermine ROI from POCs. Many enterprises lack cohesive strategies,
leading to uncoordinated efforts and limited scalability. The shortage of skilled talent further
complicates adoption, as internal teams often require reskilling or external expertise to implement
AI solutions effectively.
Externally, providers face hurdles like gaining client buy-in, ensuring compliance with evolving
regulations and addressing rigid contractual terms that do not accommodate rapidly advancing
technologies. High AI infrastructure costs and cautious client approaches to experimental investments
also limit adoption speed. Most technology services CXOs want to be agile in assessing opportunity
areas and developing their GenAI market propositions. This includes building new products and services
on the one hand and developing go-to-market roadmaps on the other. In addition, customers’ risk
appetites would continue to vary when it comes to GenAI adoption – while some have already invested in
the technology, most others are waiting for it to mature.
To overcome these challenges, technology service providers should:
Reinvent service portfolios: Develop scalable, ready-to-deploy GenAI solutions to streamline
adoption for clients
Focus on high-impact use cases: Build sector-specific applications with measurable outcomes
to deliver tangible value
Adopt outcome-based contracts: Tie client engagements to productivity gains and other
quantifiable results to demonstrate impact
In conclusion, GenAI has fast become a top strategic priority for technology services companies. The
new delivery models, talent priorities and skilling programs would continue to be debated and evolve.

The AIdea of India: 2025 89

vi
Gears of change:
Auto and mobility

G
GenAI is transforming the sector enAI is reshaping the automotive and
across design, manufacturing, mobility sector, driving innovation across
the value chain in a sector traditionally
autonomous driving, and customer defined by mechanical innovation. The advent of
engagement, enabling a shift from this technology has necessitated a re-evaluation of
product-centric to service-oriented long-standing practices, leading to advancements in
models manufacturing, system engineering and customer
experiences. Key areas where GenAI is making
Innovative business models, an impact include product development, process
including ride-sharing and fleet optimization, testing and validation, customer
management are emerging, engagement, automated driving and sustainability
initiatives. Advanced technologies like autonomous
helping organizations transition driving and real-time analytics, are central to this
to sustainable, service-driven evolution, reshaping the industry’s landscape.
operations Emerging business models like ride-sharing
and MaaS optimize fleet management and
Indian auto majors and start-ups ride-matching while facilitating the industry’s
are leading the charge in GenAI transition from product-based to
innovation leveraging technology service-centric approaches.
to streamline processes and The evolution of autonomous vehicle (AV)
enhance customer experience technology is accelerating with the emergence of
AV 2.0, defined by unified AI models capable of
GenAI is projected to handling vehicle operations such as perception,
enhance productivity by 30% to planning and control. Companies like Tesla
are driving this evolution with innovations like
32% by 2030 across key roles in
the Full Self-Driving (FSD) 12 software, which
this sector facilitates end-to-end neural network control. This
breakthrough is expected to be a catalyst for
robo-taxi deployment in 2025 and the proliferation
of other autonomous vehicle models.

90 The AIdea of India: 2025

Start-ups are also leveraging AI to revolutionize Waabi integrates Nvidia DRIVE OS for AI-defined
autonomous driving. For instance, London-based vehicle operations. These developments are laying
Wayve utilizes Nvidia Blackwell GPU architecture to the foundation for a future where fully autonomous
develop advanced AV systems, while Toronto-based vehicles can transform market dynamics.

Question Question
What has been your experience with What is your organization’s overall
GenAI POC engagements? inclination to invest in GenAI?
57% respondents have initiated POCs; with 14% running
in production

14% 5%

43%

38%

No response for ‘POCs completed - successful POCs lined up for

production’
Source : EY India C-suite GenAI Survey 2024 Source : EY India C-suite GenAI Survey 2024

The AIdea of India: 2025 91

Automotive and mobility value chain
The integration of GenAI in the automotive industry presents opportunities for innovation across the value
chain, from manufacturing to customer experiences.

Manufacturing Design and Sustainability Customer Autonomous Supply chain

innovation engineering experience vehicles planning and
(sales/after execution
Shopfloor Generative Energy sales) and Simulation
automation design management operations and testing Production
by identifying network design
Predictive patterns and
Material Auto pilot for Data
maintenance generating
optimization sales control augmentation, Cost/route
solutions
to improve path and planning
Quality and planning,
sustainability Market
compliance Virtual decision making
automation prototyping research
Warehouse
twin
automation
Demand Software and network
forecasting Virtual development
Contact
and regulatory simulation (autonomous
center
and testing Inventory
compliance automation navigation,
management
management Advanced
Driver
Marketing Assistance
Warranty claims
and campaign Systems or
and insurance
automation ADAS, NLP)
rate analysis

Global start-ups are at the forefront of this transformation, leveraging GenAI to deliver
cutting-edge solutions:

■ Skaivision: A GenAI-powered platform that ■ Artificient Mobility Intelligence: Offers smart

analyzes automotive dealership video footage mobility solutions using computer vision for
to enhance operational transparency, validate real-time driving insights, including traffic sign
performance, and optimize sales, service and detection, road risk assessment, and collision
security processes. The technology supports prevention. Its advanced features, like digital
data-driven decisions to improve efficiency, twin-driver profiling and crash detection,
profitability, and customer experience. enhance safety and reduce insurance
claim costs.
■ Ambro: Leverages AI to automate vehicle
damage detection and claim cost estimation, ■ Hello Llama: Developed AI-driven software for
streamlining claims processing with features micromobility safety, featuring the Active Safety
like real-time assistance, fraud detection, and Halo system. It enhances rider safety with
integration with automotive part catalogs. emergency braking, collision avoidance, traffic
sign detection, and parking assistance, catering
■ Brego: Offers an AI-powered platform for
to drivers and the automotive industry.
real-time automotive valuations and analytics,
supporting inventory management, risk These start-ups demonstrate how technology
assessment, and lending decisions. It serves is enabling significant advancements in safety,
various vehicle types and helps insurers, operational efficiency, and customer service.
dealers, and lenders with fraud reduction,
accurate valuations, and confident approvals.

92 The AIdea of India: 2025

Case studies: GenAI transforming Question

Indian auto majors Where has AI been impactful?

(Mark all that apply)

Leading Indian automotive companies are AI has been most impactful in driving customer
leveraging GenAI to streamline processes and satisfaction
enhance customer experience.
One four wheeler OEM has successfully integrated
GenAI into backend operations, enabling workforce
empowerment and reducing dependency on direct
customer interactions. Its AI-powered maintenance
bot assists shop floor workers by providing
step-by-step solutions in multiple languages for
complex industrial machinery issues. This approach
has minimized machine downtime and reliance on
specialized technicians while boosting
worker confidence.
On the customer support front, it uses GenAI to
generate comprehensive summaries of customer
interactions, reducing response times and
enhancing satisfaction by offering Source : EY India C-suite GenAI Survey 2024
tailored assistance.
The company has taken a significant step into the
mainstream by establishing a dedicated AI division Z-focused “Xoom,” AI drove significant lead
to enhance business outcomes and drive success recovery and customer engagement. Results
through AI. include a 33% increase in lead conversion rates,
A large two wheeler OEM employs GenAI to with over 5 million customers interacting with AI-
enhance lead management, sales pipelines, and driven systems.
customer engagement. Key solutions include
conversational AI for 24/7 customer interaction,
targeted lead conversion campaigns, and AI GenAI’s impact on productivity
visualization platforms for immersive customer
experiences. During campaigns like the Gen
Indian automotive sector, contributing 7.1% to the
national GDP and employing over 19 million people,
Question
is expected to undergo a significant transformation
driven by GenAI. According to EY India’s study of
Across organizations, select functions / over 2,300 roles in the sector, GenAI is projected
departments being prioritized for GenAI to deliver a productivity impact of 30% to 32% by
implementation (Mark all that apply) 2030, positioning the industry for growth and
operational transformation.
Marketing
57%
48%
GenAI
Customer service

is making
Sales and channel management
33% an impact in
several key areas, including product
Supply chain
19% development, process optimization,
testing and validation, automated driving
R&D
19% and sustainability initiatives
Source : EY India C-suite GenAI Survey 2024

The AIdea of India: 2025 93

94 The AIdea of India: 2025
GenAI impact on productivity across automobile value chain

Size of bubble represents the relative headcount in the business process for the sector

Source: EY India jobs study

While GenAI has the potential to influence a wide

range of business processes, certain functions /
tasks stand out due to their significant automation Challenges and future
potential: roadmap
01 Functions with highest potential for
productivity impact include sales
and customer service (37% to 39%),
Despite its advantages, the automotive
industry faces challenges in GenAI adoption,
including technical, regulatory, ethical and
production and assembly (35% to 37%) and market barriers. Compliance and ethical
maintenance (34% to 36%) which together considerations necessitate a balanced
can account for 35% to 40% of productivity approach that combines innovation with
improvement. responsibility. Collaboration among AI

02 Functions like quality assurance and control

along with marketing are also expected to
experts, automotive engineers, regulators,
and ethicists is essential to address
these challenges.
drive significant productivity improvement.
Many leading companies are making
03 Innovations in these functions are
reshaping the automotive value chain,
enabling companies to achieve cost
substantial investments in GenAI,
with dedicated budgets and focused
efficiency and operational excellence. implementation strategies. Key priorities
include areas such as manufacturing,
customer experience, and autonomous
driving. By investing in AI and focusing on
augmenting human capabilities, companies
can position themselves to lead the next
phase of industry transformation.

The AIdea of India: 2025 95

vii
Powering efficiency and
sustainability with GenAI:
Industrials and Energy

T
For the energy, power and he integration of GenAI is reshaping the
utilities sectors, GenAI can industrials and energy sectors in India,
including energy, power and utilities, oil and
improve grid stability, renewable gas, metals and mining and industrials. These
energy integration, thermal sectors are leveraging AI to address long standing
plant efficiency, and carbon challenges, enhance efficiency and drive innovation
management, enhancing cost across their value chains. From predictive
savings and sustainability maintenance to optimizing complex processes,
GenAI is providing transformative solutions while
For the oil and gas sector, GenAI enabling sustainability and improving
can optimize exploration, refining, decision-making. Below is a detailed exploration
of the impact of GenAI on these industries,
logistics, and retail with tools highlighting key use cases and transformative
for demand forecasting, applications.
pipeline monitoring, and

Industrial
sustainability initiatives
GenAI can drive smarter operations sectors
for the metals and mining sector, leverage
with advanced geological AI to address longstanding challenges,
analysis, supply chain optimization enhance efficiency and drive innovation
and emissions reduction for across their value chains
greater efficiency
In Industrials, GenAI can enhance
productivity through predictive
maintenance, generative design,
and AI-enabled safety, transforming
manufacturing and supply chains

96 The AIdea of India: 2025

Question Question
What has been your experience with What is your organization’s overall
GenAI POC engagements? inclination to invest in GenAI?
66% respondents have initiated POCs 17% 49%
17%
66%

17% 49%
are exploring this
actively with
budgets allocated
17%

17%

No responses for ‘GenAI running, in production’ and ‘POCs No responses for ‘Bullish and already investing’
completed - successful POCs lined up for production’
Source : EY India C-suite GenAI Survey 2024
Source : EY India C-suite GenAI Survey 2024

Energy, power, and utilities

Globally, the energy sector is leveraging advancements like precise forecasting, Advanced Metering
Infrastructure (AMI) and integrated management systems (ADMS and DERMS) to enhance efficiency and
reliability, with GenAI playing a pivotal role in ensuring data accuracy. In India, energy and power utilities are

The AIdea of India: 2025 97

increasingly adopting digital and AI technologies Renewable energy management: The
through either ‘Driver’ or ‘Driven’ models. ‘Driver’ renewable energy landscape benefits
utilities proactively craft AI strategies with strong significantly from GenAI. By analyzing
enterprise visions, investing in technology, talent weather patterns, grid conditions and
and data governance, while ‘Driven’ utilities focus market demand, GenAI tools identify
on using AI for specific operational challenges. optimal generation sites and provide
GenAI enhances customer engagement through accurate forecasts. For instance, short-term
personalized services and chatbots while also forecasts help maintain grid stability, while
being explored for grid optimization and asset long-term predictions assist in purchase
maintenance to reduce risks and costs, enabling planning and cost optimization. These
better energy management and fostering capabilities ensure a seamless integration
innovation for sustainable energy transitions. of renewable energy sources into
existing grids.
Investment and adoption of AI/ML vary across
companies, with mature enterprises typically Carbon management and trading: As
allocating 0.4% to 0.7% of their revenues to companies transition to low-carbon
AI/ML initiatives and achieving an ROI of operations, GenAI serves as a strategic
around 2% of annual revenues . These investments advisor for planning decarbonization
have driven value in four primary areas: operational investments. It also aids energy trading
cost reduction through AI/ML-enabled process decisions, optimizing production schedules
automation, revenue growth via differentiated and ensuring compliance with carbon
pricing strategies, improved management of trading regulations.
demand-supply variability through accurate
The adoption of GenAI in energy and utilities is not
forecasts and personalized consumer engagement
only addressing operational challenges but also
to enhance customer experience. Together, these
paving the way for sustainable growth
advancements position AI as a critical enabler
and innovation.
of efficiency, innovation, and profitability
in this sector.
The deployment of GenAI in this sector has Case study: Power demand
enabled significant improvements across various
operational and strategic areas: forecasts
Thermal plant optimization: GenAI tools
enhance combustion efficiency by analyzing Punjab State Power Corporation Limited (PSPCL)
operational inputs such as temperature, has set up an AI-ML PMU (Project Management
humidity and gas flow in real time. This Unit) as part of the Government of Punjab’s
optimization leads to reduced fuel costs, initiative to adopt best practices and develop
improved emissions compliance and a AI-based solutions for state-wide improvement
smaller carbon footprint. By integrating across sectors, including power and utilities. A key
real-time monitoring with predictive solution developed is an AI-based Load Forecast
analytics, companies can ensure that their model, which integrates 2-3 years of historical load
plants operate at peak efficiency. data from all districts to provide 24-hour ahead
forecasts at 15-minute intervals. With a Mean
Maintenance enhancement: Traditional Average Precision Error (MAPE) of 3% and a
time-based maintenance often results in self-correcting model, the solution delivers accurate
either under-maintenance or over-servicing and continuously improving recommendations.

Indian
of equipment, leading to operational
inefficiencies. GenAI enables
condition-based maintenance, using
advanced health indices to predict when energy
equipment requires attention. This organizations
reduces unplanned downtime, prolongs
are building active AI strategies for
equipment life and minimizes
unnecessary servicing costs. technology, talent, and data governance.
Some are using AI for specific
operational challenges

98 The AIdea of India: 2025

Oil and Gas Refining and transportation: Refineries
are benefiting from AI-driven process
optimization. By monitoring performance
The oil and gas sector, traditionally characterized metrics in real time, GenAI can identify
by complex, capital-intensive operations, is inefficiencies and recommend corrective
undergoing a significant transformation with actions to improve throughput and safety.
the integration of GenAI. From exploration and In transportation, AI-powered tools
production to refining and distribution, the sector is enhance pipeline monitoring and storage
leveraging GenAI to streamline operations, enhance management, minimizing risks such
safety, and reduce costs. As global markets as leaks and ensuring compliance with
increasingly emphasize sustainability and energy regulatory standards.
efficiency, Indian companies are also exploring
AI-driven solutions to remain competitive. Distribution and retail: GenAI optimizes
logistics by providing real-time route
GenAI applications in the oil and gas industry cover planning and tracking. Retail operations
a wide spectrum of operations: benefit from AI-driven demand
Exploration and production: Every facet forecasting, customer segmentation,
of oil and gas exploration and production and pricing strategies, enabling better
lifecycle is seeing GenAI’s impact, spanning inventory management and enhanced
seismic image enhancement, reservoir customer satisfaction.
simulation, creating optimal drilling plans Sustainability and ESG initiatives:
and in forecast of production profiles for GenAI supports companies in meeting
reservoir optimization, among others. sustainability goals by analyzing ESG
AI-driven solutions impact exploration risks and improving energy efficiency. For
timeline, time to surface, well trajectories, instance, renewable energy forecasting
production forecast, asset performance, powered by AI helps predict the generation
and hydrocarbon recovery, yielding of intermittent sources like wind and solar,
significant economic gains for operators. enabling better grid integration.
Beyond core domains, AI extends its
benefits to predictive asset maintenance, Indian oil and gas companies are leveraging these
document creation, employee training and applications to drive operational excellence.
safety agendas. Companies like IOCL and HPCL have adopted
GenAI for project analysis, while renewable energy
providers like Amplus are using AI to enhance
performance monitoring.

The AIdea of India: 2025 99

Case study: Improving retail sales Question
Across organizations, select functions /
A leading oil and gas company is implementing departments being prioritized for GenAI
retail AI and advanced analytics solutions to implementation (Mark all that apply)
improve sales performance and uplift margins
of their retail stores, using advanced product
assortment and demand forecasting.
Using AI, the ‘Product Assortment’ engine
recommends the best SKUs by analyzing sales,
customer and external data. ‘Store Segmentation’
clusters stores with similar attributes to optimize
marketing campaigns and reduce operational
costs. ‘Demand Forecasting’ incorporates
factors like seasonality and weather to provide
realistic demand projections, while ‘Customer
Segmentation’ groups customers by purchase
history and demographics for tailored marketing
strategies. These solutions enable granular demand Source : EY India C-suite GenAI Survey 2024
forecasts, improved sales and profit, enhanced
customer satisfaction, and better ROI through
reduced waste and targeted engagement.

100 The AIdea of India: 2025

Metals and Mining Question
Where has AI been impactful?
In the metals and mining sector, while AI (Mark all that apply)
applications have been widely implemented, AI has been instrumental in reducing costs
GenAI has primarily been applied in non-core
manufacturing functions to enhance productivity
and speed of execution. Its uses include automating
supplier and transporter interactions, extracting
actionable insights from historical supply chain
and operations data, and conducting root cause
analysis (RCA) for real-time incidents. The real
potential of GenAI lies in amplifying AI’s impact on
core operations by leveraging extensive structured
and unstructured data. Key opportunities include
simplifying complex scenario assessments, such as
furnace recipe optimization and rapidly validating
AI-driven recommendations against historical
data, offering companies a competitive edge in
operational efficiency and decision-making.
Source : EY India C-suite GenAI Survey 2024

GenAI applications have the capability to transform the metals and mining value chain

RM procurement and Logistics and

Mining consumption commercial

Geological Procurement Planning and logistics

analysis and ■ Source mix optimization ■ Dispatch planning vehicle
exploration availability forecasting
■ Alternate supplier/sourcing strategy
optimization procurement control tower ■ Production campaign
■ Contract evaluation for commodities scheduling
Remote sensing and maintenance, repair, and operations ■ Network optimization
on geospatial (MRO)/consumables ■ RCA on plan vs actual
platform
■ Spares requirement prediction deviations
■ Automated vendor negotiations, PO ■ Scenario assessment on
Mine strip integrated planning
processing and execution
ratio lifecycle
■ Automated transporter
simulations
Consumption negotiation
■ Recipe/blend optimization ■ Control tower
Mill processing
improvements ■ Intelligent Process Optimization/Digital
Twin Sales and commercial
■ Finished goods (FG) and WIP quality ■ Demand forecasting
Drill blast
predictions ■ Pricing strategy
optimization
Lead and haul ■ Fuel consumption optimization ■ Customer service
optimization ■ Performance RCA applications
■ Scenario assessments and simulations ■ SO processing and
execution
Operational excellence
■ Predictive maintenance
■ Shopfloor automation
■ Energy management systems

The AIdea of India: 2025 101

Global enterprises in the metals and mining sector The manufacturer derived tangible benefit of
are leveraging AI and GenAI to enhance efficiency, US$ 60 million – US$ 42 million in material
sustainability and cost-effectiveness. In metals, consumption, US$ 10 million from productivity
leading players like Alcoa, Nucor and Rusal, for improvement and US$ 8 million from operating/
instance, have used AI for predictive maintenance, production cost. In addition, GenAI implementation
logistics and distribution optimization and energy led to improved process visibility and control, single
optimization, respectively. In mining, while Rio version of truth, increased accountability and
Tinto has deployed digital twins and virtual mine ownership and assisted decision making.
modelling for planning and sustainability efforts,
BHP has leveraged AI for supply chain optimization
and predictive maintenance of assets. Such Industrials
solutions could have an impact across the value
chain, leading to 3% to 4% reduction in operating
costs and up to 10% in throughput, depending on The industrial sector GenAI is reshaping traditional
existing maturity levels. manufacturing, supply chain, and workforce
processes. With leadership increasingly focusing
In India, Tata Steel has deployed 150-plus on digital transformation, industrial firms are
algorithms to support real-time decision making deploying AI-driven tools to enhance productivity,
and improve operational efficiency in core blast reduce costs and create new business models.
furnace operations, raw material procurement
and asset maintenance through integrated GenAI applications in the industrial sector include:
control center. It has also achieved lighthouse
recognition with 10x returns realized over these
analytics investments.
01 Manufacturing optimization: GenAI tools
enable predictive maintenance, throughput
optimization and generative design. By
While companies have progressed on their AI analyzing machine performance data,
adoption journey, the biggest barriers they these tools can identify potential failures
face at present are: before they occur, minimizing downtime
and maximizing operational efficiency.
People capability and up-skilling Generative design capabilities allow
Limited affordability and availability of engineers to explore innovative solutions,
automation technologies accelerating product development cycles.

Low level of data quality, use case adoptions

02 Sales and marketing: AI-driven demand
forecasting and dealer churn prediction
enhance customer targeting and retention.
Virtual product demonstrations and
Case study: Consumption customized marketing campaigns improve
efficiency and cost control customer engagement, driving higher
conversion rates.
A large Indian steel manufacturer, with an EBITDA
of 21% and approximately 10 MTPA production
capacity, was facing high fuel costs, increased 03 Safety and workplace productivity:
GenAI supports workplace safety through
tools that detect compliance with safety
raw material consumption and costs and variable
quality due to people-driven decisions. gear regulations. Additionally, AI-powered
Solution: A digital twin was implemented to simulations and prototyping streamline
optimize material consumption, production costs design processes, reducing time-to-market
and productivity, using AI-ML algorithms for and improving product quality.
predictions and prescriptions.
The use of industrial robots powered by GenAI is
More than 120 parameters were considered to further revolutionizing factories. These robots,
link material characteristics, consumption and equipped with advanced object recognition and
process parameters coordination capabilities, are enabling more
An optimization engine was developed for adaptive and efficient production environments.
productivity and cost This not only reduces operational costs but also
enhances forecasting accuracy, enabling companies
A customized recommendation engine was to better align production with market demand.
developed for each process in the value chain

102 The AIdea of India: 2025

Case study: Optimizing sales
planning Challenges and way ahead
India’s industrial & energy sectors face
A leading Indian cement manufacturer is developing
GenAI adoption challenges such as poor
a Sales and Logistics Control Tower using AI models
data quality due to legacy systems,
to optimize sales planning, order management
cybersecurity risks, and the limited
and logistics. The Monthly Logistics Plan (MLP)
availability of AI talent for complex use
model generates destination and material-level
cases. Overcoming these hurdles requires a
dispatch plans, maximizing contribution while
strategic approach, including modernizing
detailing costs and dispatch splits. A 4x4 market
infrastructure, implementing robust data
mapping model supports branding and marketing
strategies, and fostering collaboration with
strategy, classifying markets to guide decisions on
technology providers. Establishing
maintaining market share or pursuing
dedicated AI units with C-suite oversight
incremental/aggressive growth. Demand Planning
and leveraging low-code/no-code platforms
employs a Consensus Planning approach,
can democratize GenAI adoption
combining top-down, bottom-up, statistical and
across organizations. By focusing on
macro-analysis forecasts to set precise monthly and
high-impact use cases and scaling successful
annual sales targets across zones and products.
pilots, companies can unlock GenAI’s
Additional solutions include a Dynamic Journey transformative potential to drive efficiency,
Planning tool that recommends optimal daily sustainability, and competitive advantage.
routes, objectives and counters. Pricing
strategies for new markets leverage 4x4 market
segmentation, providing visibility into wholesale/
retail price recommendations based on market
position and growth potential. These AI-driven
tools deliver improved sales planning, logistics
efficiency, strategic branding and tailored market
entry strategies, enhancing operational efficiency
and profitability.

The AIdea of India: 2025 103

viii
Creativity meets GenAI:
Media and
Entertainment

G
GenAI is reshaping how content enAI in the media and entertainment industry
is ideated, produced, and is marking a transformative moment, and
reshaping how content is created, consumed
experienced, from personalized and experienced. The race to develop GenAI
ads to immersive live events models capable of generating video intensified
in 2024, with its impact spanning digital media,
From multilingual localization to broadcasting, film, music, sports, and gaming.
AI-driven audience engagement In October 2024, Meta’s AI model, Movie Gen,
and virtual production, GenAI is unveiled clips of a baby hippo swimming in a
transforming media workflows and lagoon, a fire dancer performing on a beach, a
enhancing user experiences dog wearing a red-caped superhero outfit flying
through the sky - all created from text prompts.
With a projected 15% to 20% by Earlier in 2024, OpenAI introduced Sora, a GenAI
tool that allows users to generate photorealistic
2030 productivity impact on
videos — such as woolly mammoths trotting
India’s media sector, GenAI is through snowy meadows — by simply typing a
boosting customer engagement, sentence. In May, YouTube launched Veo, a video
streamlining content production generation model, and Imagen 3, a text-to-image
and improving monetization model aimed at supporting the creative process.
strategies

In India
The industry must address
concerns like copyright, deepfakes,
and preserving human creativity to ,GenAI’s impact
maintain trust and authenticity in is at present
AI-driven media mostly in advertising along with licensing
and distribution, with the potential to
redefine traditional production models

104 The AIdea of India: 2025

In India, GenAI’s impact is being realized across GenAI to unlock personalization, efficiency and
advertising, licensing and distribution, with innovative user experiences. AI-powered immersive
the potential to redefine traditional production experiences are redefining how users engage
models of having creative teams, big production with media and live events, especially in sports. In
houses and years of planning and execution. Key gaming, AI is introducing real-time interactivity,
players, both global and Indian, are exploring allowing for adaptive, personalised experiences.

Question Question
What has been your experience with What is your organization’s overall
GenAI POC engagements? inclination to invest in GenAI?
89% of the respondents have initiated POCs; with 22%
44%
running in production
22%
22%

~56%
indicate a strong
inclination to
invest in GenAI

11%
45%
56%

No response for ‘POCs completed - successful POCs lined up for No response for ‘Bullish and already investing’, ‘Exploring actively
production’ and budgets allocated’ and ‘No plans to invest’
Source : EY India C-suite GenAI Survey 2024 Source : EY India C-suite GenAI Survey 2024

The AIdea of India: 2025 105

Question Post-production: Al is revolutionizing post-
Across organizations, select functions / production workflows by automating tasks
departments being prioritized for GenAI such as video editing and visual effects
creation in addition to helping re-purpose
implementation (Mark all that apply)
older films to higher resolutions or even
restoring damaged footage.
Content creation and acquisition
44%
Business intelligence and analytics:
Marketing and promotions
44% GenAI can provide deep insights into
audience behaviours, helping companies
predict trends and refine strategies. GenAI
Ad sales and monetization
44% can significantly enhance Broadcast
Research Audience Council’s (BARC) data
Customer engagement/ CRM
33% analysis processes, helping broadcasters,
advertisers, and content creators make
more informed decisions.
Content scheduling and distribution
22%
Source : EY India C-suite GenAI Survey 2024 Intellectual Property (IP) and Rights
management: Automating contracts,
Emerging use cases syndication deals, and rights tracking are
some areas where GenAI can play a role,
offering tailored reports for compliance and
GenAI is driving innovation across various ensuring better IP management.
applications in the media and entertainment sector,
enhancing creativity and automating workflows: The emergence of sector-specific GenAI startups
Content personalization and localization: is revolutionizing India’s media and entertainment
GenAI is driving advancements in industry, transforming how content is created,
multilingual and personalized content, managed, localized and monetized. Start-ups like
transforming how media is created and Runway and Udio are pioneering AI-driven video
delivered. In FAST (Free Ad-Supported production, while Synesthesia and D-ID focus on
Streaming TV) channels, personalized realistic digital avatars and interactive content.
content feeds and efficient reuse of Platforms like DubDub.ai and Vitra.ai enable
licensed archive material are redefining multilingual and real-time localization, while
video streaming. The Tamil-language film Feltiv and Visual Dub AI enhance accessibility and
Kanguva used AI to recreate the hero’s dubbing. Tools like Frammer AI streamline content
voice in multiple languages for its release. repurposing, and Myelin Foundry offers
Music composer AR Rahman leveraged AI AI-powered distribution solutions. These startups
to revive the voices of legendary singers for are driving efficiency, personalization, and
a song in Lal Salaam, sparking a nuanced innovation, reshaping workflows and audience
debate about ethical usage. engagement across the sector.

Audience engagement through

Start-ups
AI-powered campaigns: Zee TV
collaborated with Bobble AI to engage with
85 million global users during the Sa Re Ga in the
Ma Pa 2024 auditions. AI-driven prompts
within chat apps guided users to audition
media
details, resulting in increased conversions and entertainment sector are driving
and seamless user experiences. efficiency, personalization and innovation
as well as reshaping workflows and
Virtual production and animation: GenAI
is enabling real-time effects, immersive audience engagement
narratives and virtual environments for
gaming, animation and visual media.

106 The AIdea of India: 2025

GenAI impact on productivity across media and entertainment value chain

Growth gains Efficiency gains

Size of bubble represents the relative headcount in the business process for the sector
High

Engage with customers

Schedule and
Ease of implementation

distribute content

Market content/ platforms

Medium

Produce content

Monetize content
Low

Low Medium High

Potential productivity impact of GenAI

Source: EY India jobs study

Question GenAI’s impact on productivity

Where has AI been impactful?
(Mark all that apply) India’s media and entertainment sector is
AI has been instrumental in reducing costs and driving experiencing a rapid evolution in consumer
customer satisfaction offerings and operational efficiencies due to GenAI.
Overall, GenAI is likely to have an impact of
15% to 20% across the sector based on the
200-plus roles analysed, with a ‘humans + GenAI’
approach likely to boost productivity, generate
better audio-visual outputs and help improve
monetization of products and services.

Customer engagement (~25%), content

production and marketing (~22%) can account for
21% to 25% of overall productivity improvement.
Content monetization and distribution functions
are also expected to see significant productivity
improvement.

Source : EY India C-suite GenAI Survey 2024

The AIdea of India: 2025 107

108 The AIdea of India: 2025
Challenges and way ahead
The media ecosystem must ensure authenticity and quality of content consumed by users. Critical
concerns like misinformation, intellectual property (IP) issues, deepfakes and the role of human
creativity demand attention as AI increasingly influences content creation and consumption. Key
challenges include:
Content origin and rights: Identifying content sources and ownership of AI-generated materials
(videos, images, music, scripts, etc.) remains complex. While Blockchain offers potential, its
practical adoption is still distant
Copyright issues: AI models often train on copyrighted data, leading to disputes over rights.
For instance, Dow Jones (The Wall Street Journal) and New York Post recently sued Perplexity
for infringement. Fair compensation for rights holders is essential to address these conflicts
Preserving human creativity: The growing dominance of AI-generated content risks
overshadowing human originality. A balance must be maintained to value distinctive
human-driven creations, such as live concerts or sports, which may warrant premium pricing
Misuse of visual/audio media: AI’s capacity to produce deepfakes challenges content
authenticity. To ensure trust, responsible AI practices must embed clear indications of
AI involvement. Transparency in distinguishing AI-generated from human-created
content will be pivotal
As AI transforms media, the industry faces the dual challenge of meeting consumer demands for
genuine, trustworthy content while leveraging AI’s potential for innovation. GenAI offers opportunities
for hyper-personalized experiences, dynamic monetization, and creative advancements but requires
ethical practices, regulatory compliance, and careful oversight to preserve human creativity.
By adopting AI responsibly, the media sector can balance technological progress with trust and serve
as a testbed for GenAI’s broader applications across industries.

The AIdea of India: 2025 109

Chapter 5

Government Services:
Better, faster, smarter with GenAI

110 The AIdea of India: 2025

The AIdea of India: 2025 111
Chapter 5

Government Services:
Better, faster, smarter with GenAI

T
he economic potential and transformative
impact of GenAI are prompting governments
Governments around the world are worldwide to invest in fostering AI
increasingly using GenAI in innovation. By integrating GenAI with other
policy development, resource advanced technologies, governments are
revolutionizing public services, accelerating
allocation and public engagement socio-economic development, and enhancing
decision-making processes.
The Government of India
is using GenAI to improve
International governments
inclusion, productivity and harnessing GenAI for citizen
accessibility, majorly through services
multilingual services, policy
drafting and documentation, In the US, government agencies are using
and expanding healthcare solutions GenAI in policy drafting, decision support, and
public communication. The National Institute of
Standards and Technology (NIST) is integrating
BharatGen, an open source AI into governance frameworks to support policy
multimodal multilingual LLM, generation and predictive decision-making. In law
enforcement, AI-driven models are generating
ushers in a plethora of possible crime reports and analyzing real-time incident data
solutions in public service delivery to improve police efficiency. The Federal Emergency
Management Agency (FEMA) is leveraging AI
for disaster management by simulating potential
disaster scenarios, optimizing resource
allocation, and enhancing emergency planning,
contributing to a proactive approach in
managing risks and resources.

Singapore is positioning itself as a global leader in

digital and AI integration through its Smart Nation
Initiative. GenAI is enabling authorities to simulate
future developments in urban planning and
visualize the impact of policies on infrastructure,
housing, and the environment. This citizen-centric
approach is reducing inefficiencies in public
services and leading to more effective solutions.
AI is also accelerating the policy-making process
by generating multiple drafts, allowing authorities

112 The AIdea of India: 2025

to analyze and make data-driven decisions swiftly. and enhancing the patient experience with
This process is reducing the time required to personalized, data-driven healthcare solutions.
draft policies by up to 30%, enabling Singapore to The UK government is also leveraging GenAI in
respond quickly to emerging national challenges. policy drafting and public consultation processes.
By generating multiple policy drafts and analyzing
In the UK, the National Health Service (NHS) is insights from citizen feedback, GenAI is enabling
adopting GenAI to enhance healthcare governance. more responsive and efficient policymaking.
AI-generated reports are supporting personalized
healthcare, managing extensive electronic health Estonia, renowned for its pioneering digital
record datasets, and predicting patient outcomes governance model, e-Estonia, is integrating GenAI
based on historical data. These capabilities are across public services to strengthen its digital
helping policymakers align healthcare strategies infrastructure. The Estonian government is utilizing
with population needs. For example, NHS Trusts AI-generated content in areas such as public
are deploying GenAI to automate the generation documentation, tax audits, and welfare policies.
of patient discharge summaries, saving time for By automating processes like license generation
healthcare professionals and ensuring that patients and citizen applications, Estonia is reducing
receive accurate, timely, and comprehensive administrative burden and processing times, setting
information about their treatment plans. This a global benchmark for efficient digital governance.
initiative is improving operational efficiency

Global examples of GenAI in governance

US UK Estonia Singapore
Policy drafting Healthcare Public Urban planning
Decision support Personalized healthcare documentation, tax (simulations,
audits, and welfare impact of policies
Public communication Managing electronic health policies on infrastructure,
Law enforcement record datasets housing,
Reduced
(generate crime Predicting patient outcomes administrative environment)
reports, analysis of Improved operational workload and Policy-making
real-time incident data) efficiency (automating patient processing time Data-driven
Disaster management discharge summaries) (automation decision-making
(simulations, resource Provide personalized, data- of license
allocation, emergency driven healthcare solutions generation, citizen
planning) applications, etc.)
Policy-making
Draft policies
Analysis of citizen feedback

The AIdea of India: 2025 113

India leverages GenAI for smarter NITI Aayog, India’s policy think tank, also
uses GenAI to generate comprehensive policy
governance recommendations, simulate outcomes, and
compare scenarios, ensuring more agile and
The Indian government has recognized AI as a responsive decision-making in line with evolving
critical enabler for the Digital India initiative and is socio-economic conditions.
investing heavily in AI-driven solutions to address
governance challenges. In 2024, the country has
launched some key initiatives to integrate GenAI Healthcare personalization
into public governance, with a primary focus on
accessibility, inclusivity, and promoting Indian One of the most prominent examples of GenAI
languages. The BharatGen initiative, for example, in Indian governance is its use in healthcare
aims to create foundational AI models tailored to personalization. The Ayushman Bharat initiative,
Indian languages and cultural contexts, ensuring a one of the world’s largest healthcare schemes,
more inclusive approach to AI-driven governance. is exploring the use of GenAI-driven models to
generate health reports, assess patient risks, and
provide personalized healthcare solutions. The
Policy drafting and documentation Ayushman Bharat Digital Mission (ABDM) offers a
strong digital foundation to integrate GenAI into
In India’s vast and diverse landscape, policy healthcare as its infrastructure supports effective
formulation is complex. GenAI is helping streamline data management and interoperability, enhancing
the process by automating the drafting of policies, GenAI’s impact on coordinated and
reports, and public documents, significantly patient-centric care.
reducing the time and effort required. Ministries,
such as the Ministry of Electronics and Information By generating localized and personalized content
Technology (MeitY), use AI tools to generate for users, GenAI is making government services
reports on digital governance, cybersecurity, and more accessible, efficient, and user-friendly. The
e-commerce. GenAI can quickly create drafts of government plans to further expand this model
policies, such as on renewable energy, by drawing to cover more languages and integrate AI into
from existing frameworks and global best practices, additional services like education, agriculture, and
speeding up the review and approval process, and social welfare.
enabling faster government responses.

114 The AIdea of India: 2025

BharatGen generates personalized responses, adapts to the
user’s language preferences, and interacts with
citizens in real-time, providing a seamless and
Launched on 1 October 2024, BharatGen is user-friendly experience. This case study highlights
India’s first government-supported initiative to the role of GenAI in making public services more
develop a multimodal LLM. Aimed at improving inclusive and accessible.
public service delivery and citizen engagement,
BharatGen focuses on creating foundational models
in language, speech, and computer vision that Bigger role for GenAI in public
produce high-quality text and multimodal content governance
across various Indian languages.
Spearheaded by IIT Bombay under the National The adoption of GenAI in governance brings
Mission on Interdisciplinary Cyber-Physical Systems multiple benefits, enhancing the quality, speed,
(NM-ICPS) of the Department of Science and and efficiency of public services. As governments
Technology (DST), BharatGen is designed as a handle vast volumes of data, AI models can
public good. Distinctive features of generate insights, documents, and possible
BharatGen include: solutions that would otherwise take months
to develop manually. Furthermore, AI-driven
Multilingual and multimodal models:
personalization extends to digital governance,
BharatGen’s models cover both text and
where citizens interact with public services through
speech, reflecting India’s complex
apps, websites, and other platforms. GenAI can
linguistic landscape.
generate personalized responses in real-time,
Bhartiya dataset development: Unlike improving service delivery and making interactions
models that rely on international data, with government systems more intuitive
BharatGen focuses on collecting and and user centric.
curating India-centric data, accurately
representing the country’s diverse
languages, dialects, and cultural contexts. Cost and resource optimization
Open-source platform: BharatGen
Governments, particularly in developing countries,
will be accessible to various sectors,
often face resource constraints. GenAI enables cost
leading to more transparency and
and resource optimization by automating routine
collaborative growth.
tasks, reducing manual errors, and improving
Support for a national AI ecosystem: process efficiency. By generating reports, analysing
BharatGen’s open-source nature supports datasets, and predicting outcomes, AI tools
a GenAI research ecosystem, benefiting allow government employees to focus on more
government, private, educational, and strategic tasks, such as policymaking and public
research institutions. consultation. For example, in India’s Smart Cities
Mission, AI tools analyse vast amounts of data to
In two years, BharatGen aims to contribute
generate insights on resource management, such
significantly to cultural preservation, inclusive
as optimizing water usage, energy consumption,
technology, and data sovereignty, strengthening
and traffic flow. This not only reduces operational
India’s control over its digital resources
costs but also improves the quality of life for
and narrative.
citizens by making cities more sustainable
and resilient.
UMANG app and multilingual AI As GenAI adoption increases through a variety of
applications and strategies, it is becoming a vital
India’s UMANG (Unified Mobile Application for tool in modern governance, enabling governments
New-age Governance) app serves as a one-stop to better anticipate needs, enhance service
platform for hundreds of government services, delivery, and engage with citizens more effectively.
ranging from paying utility bills to accessing The continued adoption of GenAI in governance
health records. With over 50 million users, the app promises to result in innovation, drive inclusivity,
leverages GenAI to offer multilingual support in and reshape public service delivery
13 languages, ensuring accessibility for a across the globe.
linguistically diverse population. The AI backend

The AIdea of India: 2025 115

Chapter 6

Policy agenda for India

116 The AIdea of India: 2025
The AIdea of India: 2025 117
Chapter 6

Policy agenda for India

AI policy landscape: A global

Globally, policymakers recognize the and Indian perspective
need to build global consensus across
AI is advancing rapidly, drawing significant
jurisdictions on regulating AI attention from policymakers worldwide.
Governments aim to encourage AI adoption in both
Development and deployment the economy and public services while ensuring
responsible usage. Approaches to AI governance
of AI applications in key socio-
vary across jurisdictions, reflecting different
economic sectors, akin to UPI, can national priorities. This chapter provides a global
drive innovation and private sector overview and contextualizes developments
for India.
participation. Creating scalable
applications in sectors such as
healthcare, agriculture, and education Global overview: Policy and
will foster a vibrant AI ecosystem regulatory developments

Fast track implementation of Policymakers worldwide are focused on balancing

innovation with managing risks. Efforts aim to
enhanced compute infrastructure,
ensure AI systems foster economic growth while
including deploying GPUs and addressing potential downsides. Key
empanelling cloud AI service developments include:
providers. Increased investments can
European Union:
be considered and inspiration taken
The EU AI Act
from jurisdictions like the US and
The EU AI Act represents one of the more
Singapore can help meet the growing comprehensive AI-specific regulatory framework to
demands of the AI ecosystem date. Key features include:
■ Risk management: A preemptive approach with
oversight proportional to risk tiers
■ Human oversight: Mandates for control over
high-risk systems.
■ Transparency: Requirements for traceability
and explainability in AI interactions.
■ Safety and robustness: Measures
addressing accuracy, cybersecurity, and
proportional safeguards.

118 The AIdea of India: 2025

The Act’s objectives include ensuring AI safety, ■ Safety assessment requirements: Safety
fostering legal certainty to encourage innovation, assessments should be carried out in-house or
enhancing enforcement of fundamental rights, by an entrusted third-party.
and creating a unified market for trustworthy AI. Global cooperation is increasingly prioritized to
address AI’s cross-border challenges. Initiatives
include a network of AI safety institutes and
United States forums like the Partnership for AI, fostering shared
research, evaluation, and governance practices.
The US leads global AI innovation while
adopting a cautious regulatory approach. Key
developments include: AI policy in India: IndiaAI Mission
■ DEFIANCE Act (2024): Targets misuse of and strategic frameworks
intimate digital forgeries with provisions for
damages and privacy safeguards. India’s AI policy landscape includes a mix of
■ AI leadership memorandum: Guides federal strategic initiatives, regulations, and advisories
AI adoption for national security, emphasizing designed to harness AI for economic and societal
safety, privacy, and responsible use. benefit. Key components include:

■ California AI laws: A set of comprehensive ■ IndiaAI Mission: A flagship program focused on

regulations that focus on addressing risks like positioning India as a global AI leader.
deepfakes, privacy, and education, aiming to ■ Legal frameworks: The Digital Personal Data
balance innovation with risk management. Protection Act (2023) and Responsible AI
Details presented in Annexure. Principles outline ethical AI development.
■ Policy guidelines: MeiTY advisories and NITI
Aayog’s recommendations focus on fostering
China: innovation while addressing risks.
GenAI safety requirements ■ Supporting ecosystem: Complementary
China’s regulations focus on safety assessments regulations like IT Rules ensure a holistic
and proportional risk measures for GenAI. Key approach to AI governance.
provisions include: India’s policies aim to balance innovation with
■ Corpus safety: Before using a corpus in a inclusivity and safety, ensuring AI supports national
training dataset, conduct safety assessments priorities and global leadership aspirations.
with at least 4000 corpora, ensure licensing

India’s
agreements for open-source or commercial
corpora, and maintain collection records for
self-collected datasets. policies aim to
■ Model safety: Technical measures should be balance innovation
used implemented to enhance the relevance, with inclusivity and safety, ensuring AI
reliability, and validity of AI-generated content
for its users. supports national priorities and global
leadership aspirations
■ Safety requirements: The necessity,
applicability, and safety of GenAI in various
fields must be demonstrated, with appropriate
protective measures for critical infrastructures,
and disclosure of service limitations, model
information, and personal data usage for
interactive interfaces.

The AIdea of India: 2025 119

Policy and regulatory environment concerning AI in India

Policy/ regulation Agenda/ focus

In addition to advisories issued in 2023 for intermediaries and platforms, the Ministry
issued a fresh advisory on 15 March 2024 for due diligence:
■ Users should not be able to host, display, upload and share unlawful content
MeitY advisories ■ The AI platform/software should not promote biases
■ The users should be made aware that they are working with an AI software
(can be a pop-up mechanism)
■ Synthetic content created using AI should be appropriately labelled as AI generated

The development and deployment of AI systems depends on the combination of:

■ Compute capacity
■ Foundational models
■ Datasets
■ Application development
IndiaAI Mission
■ Future skills
■ Start-up financing
■ Safe and trusted AI
This mission-based approach adopted is to develop a strategy to bridge the gaps in the
existing AI ecosystem across all identified pillars.

120 The AIdea of India: 2025

IndiaAI Mission: The way ahead

The Cabinet has approved a specific financial commitment of INR10,371.92 crore in March 2024 to
implement the IndiaAI Mission with an overarching aim to “catalyse AI innovation through strategic programs
and partnerships across the public and private sectors”. There are various initiatives under the Mission’s seven
mission pillars as below.

This pillar focuses on creating a scalable AI computing ecosystem with over 10,000 GPUs
IndiaAI through public-private partnerships to meet the demands of India's growing AI start-ups and
Compute
research community. An AI marketplace will offer AI as a service and pre- trained models,
Capacity
serving as a central hub for essential AI resources.

IndiaAI This centre is dedicated to developing and deploying indigenous Large Multimodal Models
Innovation
(LMMs) and domain-specific foundational models in critical sectors.
Centre

IndiaAI This platform aims to streamline access to high-quality non-personal datasets for AI
Datasets innovation. A unified data platform will provide seamless access to these datasets for Indian
Platform start-ups and researchers.

IndiaAI This initiative promotes AI applications in critical sectors by addressing problem statements
Application from Central Ministries, State Departments, and other institutions. It focuses on developing,
Development scaling, and promoting the adoption of impactful AI solutions with the potential for large-scale
Initiative socio-economic transformation.

IndiaAI This pillar aims to reduce barriers to entry into AI programs by increasing AI courses at the
Future undergraduate, master’s, and Ph.D. levels. Data and AI Labs will be established in Tier 2 and
Skills Tier 3 cities to offer foundational courses.

IndiaAI
This pillar supports and accelerates deep-tech AI start-ups by providing streamlined access to
Startup
Financing funding for innovative AI projects.

This pillar focuses on ensuring the responsible development, deployment, and adoption of AI
Safe and by implementing Responsible AI projects, developing indigenous tools and frameworks,
Trusted AI
self-assessment checklists for innovators, and other guidelines and governance frameworks.

Policy recommendations for the Indian AI ecosystem

Enhancing access to high-quality Recommendation

datasets The India Dataset Platform should be launched
The government is establishing the India Dataset as soon as possible, even in a beta version, to
Platform and National Data Management Office to enable early feedback from the AI ecosystem for
provide sector-specific, organized datasets with continuous improvement. The platform can be
standardized metadata and APIs for streamlined modelled after successful examples like Hugging
access. The platform will also engage with private Face, with datasets tailored to key sectors such as
entities to develop frameworks for contributing payments, healthcare, education, and agriculture,
data, allowing opportunities for monetizing readily available for AI development. Adopting
private datasets.

The AIdea of India: 2025 121

a privacy-first approach, the platform should Accelerating AI use-case
facilitate crowdsourcing for AI data needs, support
development and R&D
ecosystem collaboration (similar to ImageNet), and
function as a marketplace for private datasets. The government is prioritizing AI solutions for
key socio-economic sectors such as healthcare,
agriculture, and governance. Programs like the
India AI Innovation Challenge and IndiaAI Fellowship
Developing and enabling access to foster innovation, while a recent allocation of
compute infrastructure I₹ NR990 crore (FY24-FY28) supports three AI
Efforts are underway to improve access to compute Centers of Excellence to develop scalable solutions
resources for government agencies, start-ups, in the sectors of healthcare, agriculture and urban
and public research institutions. Initiatives include sustainability solutions.
deploying at least 10,000 GPUs and empanelling
agencies to provide AI services on the cloud, Recommendation
facilitating scalable AI research and development. The government should expedite the deployment
of AI use-case development, particularly in sectors
Recommendation like healthcare, agriculture, and education.
The government has initiated the procurement NPCI, played a pivotal role in popularizing UPI by
of GPUs. Given the growing demands in the AI providing a simple and accessible platform for
ecosystem, it is crucial to fast-track this process. digital transactions. UPI’s success has encouraged
Countries like the United States and Singapore are numerous private entities to develop their own
making substantial investments in AI infrastructure. UPI-based applications, significantly expanding the
Singapore, which has a significantly smaller digital payments ecosystem in India. Similarly, by
economy, has similarly committed over S$1 billion creating reference AI applications in key sectors,
(approximately US$740 million) over the next five the government can stimulate private sector
years to develop AI compute infrastructure, innovation and participation, leading to a diverse
talent, and industry development under its and robust AI ecosystem that addresses various
National AI Strategy 2.0. To remain competitive socio-economic challenges.
and support scalable AI initiatives, current
budgetary allocations towards AI infrastructure
might have to be reassessed and revised upwards.

122 The AIdea of India: 2025

 Promoting responsible and
trusted AI
The government recently released a report
for public consultation prepared by the
Developing
indigenous capabilities through
Subcommittee on ‘AI Governance and Guidelines
investments in local talent, research,
Development’, under its PSA chaired AI Advisory
Group. The report presents recommendations and infrastructure can not only have an
to inform the development of an AI ecosystem economic effect but can also further the
that is “trustworthy and accountable”. The development of greater sovereign AI
proposed measures include, the development capabilities
of a co-ordinated approach for the Government;
enhancement of governmental technical
capabilities; initiatives to monitor and mitigate
AI risks; drive voluntary commitments; develop Recommendation
technical standards and tools; and suggestion of Develop a standardized framework to streamline
measures to strengthen the legal and regulatory public-private AI partnerships. This framework
frameworks. should include clear review processes, standardized
SoPs, template agreements, and expedited
Recommendation approvals to formalize collaborations. Additionally,
existing facilitation centers, such as BHASKAR
Accelerate the establishment of an AI regulatory under Startup India/DPIIT, could provide specialized
sandbox, drawing from successful examples like support for startups and enterprises contributing to
the RBI’s regulatory sandbox for financial products. AI deployment in public services.
AI sandbox can help mitigate risks such as bias,
hallucination, and inaccuracy, while supporting the
creation of safety testing protocols and standards Focus on sovereign
in a controlled environment. AI capabilities
The development of AI systems (startups,
enterprises) in India have traditionally been
Streamlined mechanisms for dependent on core technologies that have been
AI public private partnerships developed in other jurisdictions. As more sensitive
ecosystem applications of AI emerge, there may risks or
considerations relating to national security and
Industry driven research and development efforts
global supply chains. Developing indigenous
for AI systems may be announced to support
capabilities through investments in local talent,
specific use-cases which have large scale economic
research, and infrastructure can not only have
implications. The India AI innovation challenge
an economic effect but can also further the
and the India AI fellowship, whereby fellows will
development of greater sovereign AI capabilities.
be supported for AI research and development,
highlights how the public is critical for AI related
research. The public at large can also be enablers
Recommendation
of access to high volumes as data which is Sovereign AI capabilities need to be established
evident through Bhashadaan which was critical in sectors of critical national strategic
for the development of Bhashini. Other initiatives importance such as space, defence, finance,
announced by the government include the etc. A programmatic approach could be adopted
announcement made by the Government of Andhra to facilitate the development of indigenous
Pradesh for the setting up of an AI university under technology, algorithms and models for all critical
the PPP model. or sensitive AI applications. Government programs
focusing on indigenous manufacturing of critical AI
related hardware may also be considered.

The AIdea of India: 2025 123

Annexures
124 The AIdea of India: 2025
The AIdea of India: 2025 125
Annexures

A
time units to estimate their volume and associated
effort. This multidimensional framework enabled a
precise understanding of the AI impact on
specific tasks.
Methodology of EY jobs study
The intersection of exposure, complementarity,
and intensity for each task produced a productivity
percentage, indicating the proportion of time or
EY undertook a comprehensive study to evaluate effort that could be saved through the adoption
the potential impact of AI, with a particular focus of GenAI. These productivity percentages were
on GenAI, on jobs across industries in India. subsequently aggregated at both the business
Our study evaluated more than 10,000 tasks, process and industry levels, enabling the analysis
across core industry operations and support to identify commonalities in processes across
functions. The methodology was underpinned industries and to compare and contrast the
by EY’s proprietary tools, frameworks, and potential impact of GenAI across different sectors.
sectoral expertise, developed over years of client A time horizon of 3-5 years was adopted to project
engagements and research. the impact of GenAI on the identified tasks.
The initial step involved mapping the value chain The findings were rigorously validated by EY’s
for each industry to identify core processes and sector-specific industry experts to ensure
support functions. Using EY’s proprietary tools and analytical robustness and accuracy.
frameworks, these processes were systematically To estimate the broader impact of productivity
broken down into progressively granular gains at economy level, the sector-specific
levels—Level 1 (L1), Level 2 (L2), and Level 3 (L3). productivity impacts were aggregated using the
Each L3 process was further detailed into a set RBI’s KLEMS database. The estimated labour
of tasks, with a level of granularity that ensured productivity gains and labour income share in
consistent time valuation across tasks. By breaking gross output for RBI KLEMS sectors were used
down tasks into granular levels, the study ensures to compute the impact of labour productivity
that the time taken for each task is consistently gains on the sectoral outputs. Gross output gains
measured and comparable. This standardization for each KLEMS sector from improved labour
allows for accurate analysis, forming a baseline to productivity were aggregated to arrive upon the
assess the impact of GenAI or other automation overall productivity gains for India’s gross output.
technologies on these tasks. KLEMS sectoral employment numbers were used
to estimate the impact of productivity gains on
Each identified task was assessed using a robust employees. These steps helped provide valuable
framework based on three critical dimensions. insights into AI-driven productivity gains at sectoral
First, Exposure, which measured the extent as well as at an economy level.
to which a task lends itself to productivity
enhancements through automation and Gen AI. This structured methodology enabled a
Second, Complementarity, which examined the granular understanding of AI’s potential across
degree of human oversight required, indicating sectors, laying the foundation for actionable
the potential for AI augmentation. Third, Intensity, insights into productivity enhancement and
which analysed the frequency of tasks in granular workforce transformation.

126 The AIdea of India: 2025

B
are consequential and relevant for all businesses
which have plan to operationalize, develop or
deploy AI systems. These laws focus on areas like
Summary of laws passed in the Privacy, AI misuse, Training data, transparency,
State of California, USA accountability and AI education which are major
considerations for the AI ecosystem at large. Since
these laws are passed in California, which is the
hub of all major technology players, these laws are
In September 2024, a series of 18 AI laws significant for all AI players globally. A summary of
was passed by California to “advance safe and these laws and their provisions are
responsible AI and protect Californians”. These laws presented below.

Name of the law Focus area Provisions

Artificial Intelligence AI Formalizes a uniform definition for AI, or artificial intelligence, in

(AB 2885) definition California law.

California Consumer Recognizes and highlights that personal information under the California
Privacy Act (AB Privacy Consumer Privacy Act (CCPA) can be stored, transmitted, recreated and
1008) shared by AI systems in various formats.

Crimes: child AI misuse

Includes content that is created digitally or through AI in the scope of the
pornography (SB and
present child pornography regulations.
1381) deepfakes

Crimes: Child AI misuse

Child pornography content altered or created using AI systems is brought
Pornography (AB and
under the scope of existing child pornography statutes.
1831) deepfakes

Prohibits the production, distribution and transmission of the digital

Use of likeness: AI misuse
replica of a deceased personality’s voice or likeness in an expressive
digital replica. (AB and
audiovisual work or sound recording without prior consent,
1836) deepfakes
except as provided.

Defending
Democracy from AI misuse Provides injunctive relief in cases where deceptive material
Deepfake Deception and (altered/created using AI systems) concerning elections is made available
Act of 2024 (AB deepfakes by large online platforms with at least one million California users.
2655)

Elections: Extends the period during which a committee or similar body is restricted
AI misuse
deceptive media in from intentionally disseminating advertisements or electoral materials
and
advertisements (AB that include misleading content created or altered by artificial intelligence
deepfakes
2839) from 60 days prior to an election to 120 days, among other measures.

Establishes a criminal offense for an individual who deliberately creates

and disseminates any sexually explicit depiction of an identifiable
Crimes: distribution AI misuse
individual, made in such a way that a reasonable person would assume
of intimate images and
the image is a genuine portrayal of the person shown, in situations where
(SB 926) deepfakes
the distributor is aware or ought to be aware that sharing the image can
cause emotional distress for the depicted individual.

The AIdea of India: 2025 127

128 The AIdea of India: 2025
Name of the law Focus area Provisions

Sexually explicit
AI misuse and Mandates social media platforms to introduce mechanisms for
digital images (SB
deepfakes reporting and removing “sexually explicit digital identity theft.”
981)

Public schools: Mandates the Superintendent of Public Instruction (SPI) to set up a

artificial intelligence AI literacy and working group in order to observe how artificial intelligence (AI) and
working group (SB education other forms of similarly advanced technologies are presently being
1288) deployed in education.

Pupil instruction:
media literacy:
Mandates the Instructional Quality Commission (IQC) to potentially
artificial intelligence
AI literacy and include AI literacy programmes in the mathematics, science,
literacy: curriculum
education and history-social science curriculum frameworks and
frameworks:
instructional materials.
instructional
materials. (AB 2876)

Generative artificial
Training
intelligence: training Mandates that AI developers provide relevant information on the
datasets and
data transparency data used to train the AI system or service on their websites.
transparency
(AB 2013)

Political Reform Act

Mandates committees that create, publish, or distribute a political
of 1974: political AI use in
advertisement that contains any image, audio, or video that is
advertisements: political
generated or substantially altered using AI to include a disclosure
artificial intelligence advertisements
identifying the content as AI-generated.
(AB 2355)

Health care services: Mandates specified health care professionals to clarify deployment of
Transparency
artificial intelligence GenAI if it is used to generate communications to a patient regarding
in AI use
(AB 3030) their clinical information.

Contracts Stipulates that a contract for the provision of personal or

against public professional services that includes a clause permitting the utilization
policy: personal Transparency of a digital imitation of a person’s voice or image is not legally
or professional in AI use binding unless it encompasses a detailed explanation of the planned
services: digital applications of the imitation and the person is represented by an
replicas. (AB 2602) attorney or a labor union, as detailed.

California AI Mandates the developers of covered GenAI systems to both incorporate

Transparency
Transparency Act provenance disclosures in the original content generated by them and
in AI use
(SB 942) provide tools to detect GenAI content generated by their systems.

Sets forth conditions for health plans and insurers regarding

Fair, their employment of AI in utilization review and management
Health care
responsible and determinations, mandating that AI, algorithms, or any software
coverage: utilization
transparent use employed must rely on a patient’s medical or clinical history and the
review (SB 1120)
of AI systems specific clinical context provided by the treating provider, and must
not replace the decision-making of healthcare providers.

Mandates CDT to update report for the Governor as called for in

Generative Artificial Executive Order N-12-23, related to the procurement and deployment
Intelligence Accountability of GenAI by the state; requires OES to carry out a risk analysis of
accountability Act. in AI use potential threats posed by the use of GenAI to California’s critical
(SB 896) infrastructure (w/high-level summary to Legislature). It also mandates
that the use of GenAI for state communications be stated clearly.

The AIdea of India: 2025 129

 Core team Healthcare Other EY AI Leaders
Acknowledgements ■
■
■
Rajiv Memani
Mahesh Makhija
Gaurav Goyal
■ Kaivaan Movdawalla
■ Srimayee Chakraborty
■ Ankur Dhandharia
■ Eisha Anand
■
■
■
Prashant Garg
Santosh Tiwari
Sumanta Kar
■ Hari Balaji ■ Vijay Shankar
■ Nisha Sharma
■ Aditi Gokhale ■ Srabati Nandy ■ Burgess Cooper
■ Murali Rao
EY jobs study Life sciences
■ Neha Sharma ■ Suresh Subramanian Brand, marketing &
■ Rajnish Gupta ■ N Balaji communications
■ Divya Chandrasekhar ■ Jerin Verghese
Technology services ■ Sharon Dias
■ Divya Maria Joy
■ Nitin Bhatt ■ Isha Handa
■ Shubham Agarwal
■ Abhinav Johri ■ Pallavi Janakiraman
■ Abhishek Trivedi
■ Asmita Mishra
Auto, Industrials and Editorial
■ Himanshu Sharma
Energy
■ Pragati Gupta ■ Prosenjit Datta
■ Neville Dumasia
■ Harshita Sinha ■ Kaveri Nandan
■ Vinay Raghunath
■ Sanjali Chandak ■ Neel Goyal ■ Radhika KTP
■ Gagan Sharma ■ Gaurav Moda ■ Vikram D. Choudhury
■ Ritik Goel ■ Vikram Mehta ■ Sushmita Yadav
■ Rishi Dewan ■ Niharika Thakur ■ Shweta Sharma
■ Pratik Shah

Sector specialists Design, layout and

Media and
Entertainment infographics
Financial Services
■ Ashish Pherwani ■ Jaspal Singh
■ Pratik Shah
■ Raghav Anand ■ Ritika Saini
■ Rohit Pandharkar
■ Sheetal Disale ■ Shubh Mittal ■ Tobit Pinto

Retail, consumer and Government

e-commerce ■ Rakesh Kaul Punjabi
■ Ashish Nanda ■ Anurag Dua
■ Angshuman
Bhattacharya
Policy
■ Digvijay Ghosh
■ Rajnish Gupta
■ Ankan De
■ Oleina Bhattacharya

130 The AIdea of India: 2025

The AIdea of India: 2025 131
Ernst & Young LLP
EY | Building a better working world
EY is building a better working world by creating new value for clients,
people, society and the planet, while building trust in capital markets.

Enabled by data, AI and advanced technology, EY teams help clients

shape the future with confidence and develop answers for the most
pressing issues of today and tomorrow.

EY teams work across a full spectrum of services in assurance,

consulting, tax, strategy and transactions. Fueled by sector insights, a
globally connected, multi-disciplinary network and diverse ecosystem
partners, EY teams can provide services in more than 150 countries and
territories.

All in to shape the future with confidence.

EY refers to the global organization, and may refer to one or more, of
the member firms of Ernst & Young Global Limited, each of which is
a separate legal entity. Ernst & Young Global Limited, a UK company
limited by guarantee, does not provide services to clients. Information
about how EY collects and uses personal data and a description of the
rights individuals have under data protection legislation are available via
ey.com/privacy. EYG member firms do not practice law where prohibited
by local laws. For more information about our organization, please visit
ey.com.
Ernst & Young LLP is one of the Indian client serving member firms of EYGM Limited.
For more information about our organization, please visit www.ey.com/en_in.

Ernst & Young LLP is a Limited Liability Partnership, registered under the Limited
Liability Partnership Act, 2008 in India, having its registered office at Ground Floor,
Plot No. 67, Institutional Area, Sector - 44, Gurugram - 122 003, Haryana, India.

© 2025 Ernst & Young LLP. Published in India. All Rights Reserved.

EYIN2501-009
ED None

This publication contains information in summary form and is therefore intended

for general guidance only. It is not intended to be a substitute for detailed research
or the exercise of professional judgment. Neither EYGM Limited nor any other
member of the global Ernst & Young organization can accept any responsibility for
loss occasioned to any person acting or refraining from action as a result of any
material in this publication. On any specific matter, reference should be made to the
appropriate advisor.

RS2

ey.com/en_in

@EY_India EY EY India EY Careers India @ey_indiacareers

132 The AIdea of India: 2025