Arne Rossmann

→ What if your AI could think, learn, and adapt like a human? The world of Retrieval Augmented Generation (RAG) is evolving rapidly - and the next generation is more powerful than ever. • Traditional RAG blends your AI model with external data retrieval. It gives your answers context and accuracy. • Self-RAG goes further. The AI queries itself, refining responses autonomously. • Agentic RAG adds agency - your AI becomes an active decision-maker, choosing how to gather and use information. • Corrective RAG introduces a feedback loop. The AI learns from mistakes and corrects itself for higher precision. • Adaptive RAG is the smartest of all. It dynamically changes retrieval strategies based on context, constantly evolving to serve your needs better. follow Devjyoti Seal for more insights

133

10 kommentarer

𝗕𝘂𝗶𝗹𝗱𝗶𝗻𝗴 𝘁𝗵𝗲 𝗗𝗶𝗴𝗶𝘁𝗮𝗹 𝗥𝗲𝗮𝘀𝗼𝗻𝗶𝗻𝗴 𝗧𝗵𝗿𝗲𝗮𝗱 𝗶𝗻 𝗘𝗻𝗴𝗶𝗻𝗲𝗲𝗿𝗶𝗻𝗴: 𝗧𝗵𝗲 𝗙𝗼𝘂𝗻𝗱𝗮𝘁𝗶𝗼𝗻 𝗳𝗼𝗿 𝗜𝗻𝗱𝘂𝘀𝘁𝗿𝗶𝗮𝗹 𝗚𝗲𝗻𝗔𝗜 Together with Dr. Dirk Alexander Molitor and colleagues at Accenture Industry X, we are developing a framework for applying GenAI and Agentic AI in engineering, building on their proven value in software development. In software engineering and SDLC, the success of LLMs is predictable: code is highly structured, follows standardized syntax, has clear semantics, and can be fully compiled and tested. In other words: 👉 Classical software problems are machine-readable and 𝒄𝒐𝒎𝒑𝒊𝒍𝒂𝒃𝒍𝒆. This makes LLMs efficient at generating, interpreting, and improving software code and iterations - with little to no preprocessing. ⚙️𝗕𝘂𝘁 𝘄𝗵𝗮𝘁 𝗵𝗮𝗽𝗽𝗲𝗻𝘀 𝘄𝗵𝗲𝗻 𝘄𝗲 𝘁𝘂𝗿𝗻 𝘁𝗼 𝗲𝗻𝗴𝗶𝗻𝗲𝗲𝗿𝗶𝗻𝗴 𝗱𝗮𝘁𝗮? Unlike software, engineering data is 𝘀𝗰𝗮𝘁𝘁𝗲𝗿𝗲𝗱 𝗮𝗰𝗿𝗼𝘀𝘀 𝗱𝗶𝘀𝗰𝗶𝗽𝗹𝗶𝗻𝗲𝘀 𝗮𝗻𝗱 𝘁𝗼𝗼𝗹𝘀. It comes in a wide range of formats - from requirement documents and SysML diagrams to CAD models and high-resolution simulations. Much of it is semi-structured, unstructured, or only partly interpretable by machines. Semantics are often implicit, domain-specific, and context-dependent. Most engineering data is not compilable and can’t be validated without connecting it to related data. 💡 𝗧𝗵𝗮𝘁’𝘀 𝘄𝗵𝘆 𝗿𝗮𝘄 𝗶𝗻𝗽𝘂𝘁 𝗶𝗻𝘁𝗼 𝗟𝗟𝗠𝘀 𝗼𝗿 𝗥𝗔𝗚𝘀 𝗮𝗹𝗼𝗻𝗲 𝗶𝘀𝗻’𝘁 𝗲𝗻𝗼𝘂𝗴𝗵 𝗶𝗻 𝗲𝗻𝗴𝗶𝗻𝗲𝗲𝗿𝗶𝗻𝗴. They require structured inputs and 𝗰𝗼𝗻𝘁𝗲𝘅𝘁𝘂𝗮𝗹 𝗮𝘄𝗮𝗿𝗲𝗻𝗲𝘀𝘀 to generate meaningful output. To make GenAI and Agentic AI truly effective in engineering, we need to transform the data, connect it across systems, and make it machine-readable. 👉 This includes structured data transformation, semantic enrichment, knowledge graph integration, feature extraction, and embedding into accessible engineering platforms - whether through APIs, agent communication protocols (MCP/A2A), or tool-level integrations. 𝗧𝗵𝗮𝘁’𝘀 𝘁𝗵𝗲 𝘃𝗶𝘀𝗶𝗼𝗻 𝘄𝗲’𝗿𝗲 𝘄𝗼𝗿𝗸𝗶𝗻𝗴 𝘁𝗼𝘄𝗮𝗿𝗱: Once engineering data is machine-readable, interpretable, and connected, 𝘄𝗲 𝗰𝗮𝗻 𝗹𝗲𝘃𝗲𝗿𝗮𝗴𝗲 𝘁𝗵𝗲 𝘀𝗮𝗺𝗲 𝗿𝗲𝗮𝘀𝗼𝗻𝗶𝗻𝗴 𝗰𝗮𝗽𝗮𝗯𝗶𝗹𝗶𝘁𝗶𝗲𝘀 𝗮𝗰𝗿𝗼𝘀𝘀 𝘁𝗵𝗲 𝗱𝗶𝗴𝗶𝘁𝗮𝗹 𝘁𝗵𝗿𝗲𝗮𝗱 𝗳𝗼𝗿 𝗽𝗵𝘆𝘀𝗶𝗰𝗮𝗹 𝗽𝗿𝗼𝗱𝘂𝗰𝘁𝘀 as is now possible in the software development life cycle. 𝗜𝗳 𝘆𝗼𝘂’𝗿𝗲 𝗶𝗻𝘁𝗲𝗿𝗲𝘀𝘁𝗲𝗱 𝗶𝗻 𝗺𝗼𝗿𝗲 𝗱𝗲𝘁𝗮𝗶𝗹𝘀 𝗼𝗿 𝘄𝗼𝘂𝗹𝗱 𝗹𝗶𝗸𝗲 𝘁𝗼 𝘀𝗵𝗮𝗽𝗲 𝘁𝗵𝗶𝘀 𝘁𝗼𝗽𝗶𝗰 𝘁𝗼𝗴𝗲𝘁𝗵𝗲𝗿, 𝗿𝗲𝗮𝗰𝗵 𝗼𝘂𝘁 𝘁𝗼 𝘂𝘀 - 𝗹𝗲𝘁’𝘀 𝘀𝗵𝗮𝗽𝗲 𝘁𝗵𝗲 𝗳𝘂𝘁𝘂𝗿𝗲 𝗼𝗳 𝗶𝗻𝗱𝘂𝘀𝘁𝗿𝗶𝗮𝗹 𝗔𝗜 𝘁𝗼𝗴𝗲𝘁𝗵𝗲𝗿. ---- Daniel Spiess | Dr. Matthias Ziegler | Liam Friel | Dr.-Ing. Tobias Guggenberger | Boris Scharinger

146

20 kommentarer

#AI #Data & #Digital #Profession: An interview podcast series with leading industry experts for advancing the AI, Data & Digital Profession, and #knowledge #sharing for #public #good. Business Architecture & Operating Model: In an recent interview (episode #4), I sat down with one of the leading architect and innovator, Dr. Terry Roach and deep dived into the topic of Business Architecture & Operating Model, which is a strategic foundation for Enterprise Scale AI Systems architecture, solutions and beyond. -> Interview Video: https://lnkd.in/gYv62sRJ Terry has extensive experience as a practitioner and leader over 30+ years, and shared critical insights around the following points for graduates and professionals. 1. Business architecture elements 2. Operating model elements 3. Business architecture and operating model integration 4. AI powered business architecture 5. AI powered operating model 6. Modernising enterprise architecture teaching & skills for upcoming professionals Sharing an important quote from Terry Roach's interview explaining the pragmatic connection between the business architecture and operating model: "The operating model will generally exist. It's there. You already have systems, you already have processes, you already have people doing stuff. What you haven't probably got is a really good structure for explaining what they all do. What the business architecture does is it gives you the format and the structure to organise the operating model knowledge into that structure". Thank you so much Terry Roach, I am grateful to you for our long standing applied research and innovation partnership over the period of last 10 years. I look forward to continue our collaboration and knowledge sharing at the School of Computer Science (Software Engineering), UTS Faculty of Engineering and IT. #AI #EnterpriseArchitecture #BusinessArchitecture #OperatingModel

81

3 kommentarer

In recent discussions with customers on AI, Responsible AI is in the spotlight. The key focus is the EU AI Act and how it impacts organizations operating in Europe. When the EU AI Act was first announced last year, many saw it as just another regulation. But forward-thinking organizations are beginning to see it as an opportunity to turn compliance challenges into a competitive advantage. Why the EU AI Act Matters: 🟩 It impacts any organization using AI that affects EU citizens, not just EU companies. 🟩 Penalties for non-compliance include fines up to €35M or 7% of global turnover. 🟩 From August 2025, providers of generative AI must comply with transparency rules, such as clearly disclosing that content has been AI-generated. At HCLTech, we enable organizations to navigate this regulation with ease, helping them transform compliance into a competitive edge. We have a team of 600 plus GRC and privacy consultants with deep knowledge of regulatory frameworks, and we deliver frameworks that are aligned with global standards like ISO, NIST and OECD.  We combine our people and process expertise in Responsible AI with cutting-edge tools to ensure compliance, innovation, and trust: Our Tools for Responsible AI 🟩 Assure AI: An LLM testing tool to evaluate AI systems for compliance and ethical risks. 🟩 TrustiFAI: A solution to manage bias and fairness in AI systems, ensuring inclusivity. 🟩 ORA Toolbox: A suite of 30+ tools for explainability and realtime monitoring of AI systems. The EU AI Act isn’t just a challenge. It is an opportunity to lead responsibly in the AI-driven world. Please read the recent whitepaper by HCLTech on the EU AI Act, and we look forward to building AI systems for our customers that inspire trust, ensure compliance, and drive growth. #EUAIACT #ResponsibleAI #HCLTech #AIRegulation #ComplianceLeadership

26

1 kommentar