When Will AI Virtual Support Agents Replace

Your IT Service Desk?

Published: 24 April 2017 ID: G00320107

Analyst(s): Chris Matchett, Rich Doheny, Kenneth Gonzalez, Magnus Revang

Hype about AI grows as consumers become familiar with virtual assistants

using conversational platforms. I&O leaders responsible for the IT service
desk are keen to exploit this to optimize IT support, but neither the
technology nor the workplace is ready to depend on virtual agents now.

Impacts
■ Business users' experience with virtual assistants for personal tasks puts pressure on I&O
leaders to meet growing expectations for the IT service desk.
■ Many business users will avoid engaging with virtual support agents, making it challenging for
I&O leaders to encourage unwilling users to engage with alternatives to human contact.
■ Virtual support agents without access to a rich source of knowledge cannot provide intelligent
responses, forcing I&O leaders to establish or improve knowledge management initiatives.
■ Current implementations of AI on IT service desks are limited to simple chatbot transactions, so
I&O leaders have no fully proven cases of virtual support agents to optimize IT support.

Recommendations
I&O leaders looking to optimize IT operations using ITSM should:

■ Determine receptivity for virtual support agents before making any investment in the technology
by observing consumer trends and through directly engaging with business users.
■ Procure VSA technology only when the benefits of efficiency and additional contact channels
outweigh any negative impact on customer satisfaction that is forecast by discussions with
business users.
■ Plan to wait for complete VSA solutions "off the shelf" to avoid developing experimental
technology in tandem with vendors. New deployments must expect to wait at least a year
before they can begin to meet business-user expectations.
 ■ Establish a foundation in knowledge management using techniques like Knowledge-Centered
Service (KCS) to build a valid and dynamic knowledge base before implementing VSA
technology.

Strategic Planning Assumption

Through 2020, 99% of artificial intelligence (AI) initiatives in IT service management (ITSM) will fail
due to the lack of an established knowledge management foundation.

Analysis
Improvements to natural-language processing (NLP) technology enable a greater role for virtual
customer assistants (VCAs) in customer service processes. By 2020, 25% of customer service and
support operations will integrate virtual customer assistant technology across engagement
channels, up from less than 2% in 2015 (see "Market Guide for Virtual Customer Assistants"). Virtual
customer assistants feature similar technology to virtual support agents (VSAs), but are designed for
a customer engagement center scenario, with integration points and user expectations that do not
fit the IT support use case for internal IT service desks. Technology and NLP alone are not sufficient
to enable the success of virtual support agents in an IT support scenario.

Page 2 of 13 Gartner, Inc. | G00320107

 Figure 1. Three Ways AI Impacts the IT Service Desk

Source: Gartner (April 2017)

Figure 1 demonstrates that AI can impact the IT service desk in three key ways. Use Table 1 to
understand the differences between these, and how they contrast with other similar technologies
outside ITSM use cases, which in turn will help infrastructure and operations (I&O) leaders evaluate
these options correctly.

Gartner, Inc. | G00320107 Page 3 of 13

 Table 1. IT Service Desk AI Technology Definitions You Need to Know

Technology Definition

Chatbots A chatbot is a type of conversational agent, a computer program designed to simulate an intelligent
conversation with one or more human users in natural language via auditory or textual methods.

Virtual A virtual assistant is a business application that simulates a conversation in order to deliver
assistants information and, if advanced, performs transactions on behalf of the user. The main difference
between a chatbot and virtual assistant is the ability to perform transactions.
Examples from outside ITSM include virtual customer assistants and virtual personal assistants.

Virtual support VSAs are virtual assistants that provide IT support and assistance in an IT service management
agents scenario, alongside the IT service desk. They pull information from knowledge management
sources and an ITSM tool to provide answers to common questions. They extend chatbot
capabilities by also taking action on behalf of the business user to do things like reset passwords,
deploy software, escalate support requests and carry out changes to restore IT services.

Artificial Technologies that learn on their own and can produce unanticipated results (see "How to Define
intelligence and Use Smart Machine Terms Effectively").
There are three key requirements for AI:
1. Adapting behavior based on experience.

2. Not being totally dependent on instructions from people (they learn on their own).

3. Being able to come up with unanticipated results.

Algorithmic IT AIOps platforms utilize big data, modern machine learning and other advanced analytics
operations technologies to directly and indirectly enhance IT operations (monitoring, automation and service
(AIOps) desk) functions with proactive, personal and dynamic insight.
Outputs related to IT service desk include trend analysis of incidents, problems and changes to
provide context and knowledge that can be leveraged by ITSM tools and support staff. This data
fuels automated decision making in VSAs.

Source: Gartner (April 2017)

Virtual support agents have been described as "smart machines." The reality is that the VSA or
chatbot products currently on the market do not utilize enough aspects of AI (learn, predict and
surprise) to be considered smart. The conversational platforms utilize NLP, but otherwise are
1
scripted solutions that do not adapt and never produce unanticipated results (barring bugs). I&O
leaders are interested in VSA, but are too focused on the novelty of hyped technology and have lost
sight of business needs and user engagement (see Figure 2).

Page 4 of 13 Gartner, Inc. | G00320107

 Figure 2. Impacts and Top Recommendations for I&O Leaders

Source: Gartner (April 2017)

Impacts and Recommendations

Business users' experience with virtual assistants for personal tasks puts pressure on I&O
leaders to meet growing expectations for the IT service desk

Virtual assistants have evolved from a concept of science fiction into real-world products that have
2
integrated themselves into people's lives. Adoption of consumer-facing virtual personal assistants
3
such as Apple Siri, Amazon Alexa, Microsoft Cortana and Android Now is increasing. People are
turning to both generalist and specialist VPAs as new interfaces to perform knowledge retrieval,
provide proactive information (such as driving directions) and complete common tasks (such as
setting reminders and appointments). User interactions with virtual assistants are also growing in
frequency. Beyond mobile devices, virtual assistants are permeating other facets of the consumer

Gartner, Inc. | G00320107 Page 5 of 13

 world, including virtual customer assistants in the B2C space (a market estimated at $450 million,
with approximately 2,500 customer deployments — see "Market Guide for Virtual Customer
Assistants"), which are being used to simulate human conversations to deliver information and
perform transactions on behalf of the business.

VSAs, therefore, enable I&O leaders to mirror users' VPA experiences outside of corporate IT while
offering additional support channels to engage with business users and automating support. This
also complements the idea of an omnichannel IT service desk, in which support shifts from a single
point of contact to multiple channels, each requiring a different scope of IT involvement (see Figure
3).

Figure 3. IT Service Desks Move From Single-Point-of-Contact to Omnichannel

Source: Gartner (April 2017)

VSAs offer an interactive user experience in accessing data or executing transactions without
directly utilizing Level 1 agents. They should switch the traditional learning paradigm — from users
learning the self-service portal, to the VSA learning about users through NLP and machine learning.
Theoretically, more advanced tools, with self-learning capabilities, also offer the promise of building
new associations based on inputs, outputs, and corresponding user feedback and scoring — in
other words, the tools shift from programmatic bots to real AI.

Currently, however, the state of chatbots and VSAs is immature. Providers offering these solutions
have yet to settle on a taxonomy; the terms "chatbots," "VSA" and "virtual assistants" are used
interchangeably, and a true AI solution has yet to emerge. Buyers are confused, and the promise is
often greater than the execution.

Page 6 of 13 Gartner, Inc. | G00320107

 As a result of the market's immaturity and current limitations in the products, I&O leaders should not
plan on replacing their traditional channels with an AI service desk within the next five years.
Instead, they should look for opportunities to add or augment channels in their omnichannel service
desk.

Recommendations:

■ Monitor developments in virtual support agent technology by reviewing vendor updates,

industry news and gartner.com to avoid developing this technology in tandem with the vendor.
New deployments must expect to wait at least a year before this deployment can begin to meet
business user expectations.
■ Be wary of vendor claims around AI capabilities — the market is fraught with hype, and a
thorough understanding of the details of chatbots, VSA and AI in general is needed for proper
due diligence.

Many business users will avoid engaging with virtual support agents, making it
challenging for I&O leaders to encourage unwilling users to engage with alternatives
to human contact
I&O leaders exploring AI technologies for IT service desk are looking at VSA products and asking
Gartner about the technology without asking the business users that would use the support
4
channel. Despite the potential of virtual support agents, cultural resistance may prove to be one of
the biggest initial inhibitors within the enterprise IT service desk. For virtual support assistants'
technical support to be accepted, they must establish trust and deliver value to the business users.

Trust is not an automatic attribute and must be earned. This brings up a number of challenges for
VSAs in their infancy phase. They must not only establish that they can deliver correct solutions to
users' issues in a consistent manner, but they must also be able to direct users to the correct
locations when the incident or request goes beyond the scope of their programmed capabilities. If
this trust cannot be established, users will default to their more traditional support channels. If those
are unavailable, their satisfaction with IT will be diminished. As with other automation initiatives,
there may be internal resistance if IT staff feels threatened by future replacement through the VSA.
This impacts the willingness of the support staff to market the VSA, and to continually develop and
validate its logic.

As with any relationship, an important element to building trust is honesty. Tell business users
upfront that they are dealing with a VSA. Attempting to deceive users raises "creep factor" issues
for some and, more significantly, undermines trust that is essential for user adoption of VSAs.

Simple avatars are fine, but attempts to use animation to make the avatar seem human will trigger
5
the "uncanny valley" effect (see Note 2). A simulation of a human face that is "just off" is off-putting
and unpleasant to interact with (see "Digital Businesses Need to Understand the Creepy Line From
Peoples' Perspective").

Gartner, Inc. | G00320107 Page 7 of 13

 But simply developing new support channels does not guarantee interest. Users typically will take
the path of least resistance or stick to paths that are comfortable for them in attaining support. As a
result, VSAs must provide added value, such as faster time to resolution, saved steps by leveraging
contextual information, more natural user experience, a new delivery mechanism (such as web chat
or more natural-language support over traditional knowledge base) or providing extended hours of
support (human-staffed desk during working hours with VSA after hours).

Even with trust and value established, some users will still prefer engaging with human contacts or
through channels not supported by the VSA (for example, phone or walk-up support). A Gartner
6
research study revealed that just 3% of organizations surveyed have deployed chatbots or VSAs.
Phone contact is still the predominant channel for volume into the service desk. Avoidance of
automated systems such as self-service is a cultural factor that cannot be ignored. The same study
revealed that 64% of respondents identified business culture/apathy as one of top barriers to using
IT self-service in their organizations.

Recommendations:

■ Determine business-user interest for virtual support assistants — by observing consumer trends
and through direct engagement including surveys, focus groups and product demos with the
users — before making a major investment in the technology.
■ Invest in virtual support agent technology only when the benefits of efficiency and additional
contact channels outweigh any negative impact on customer satisfaction.
■ Build trust by telling business users that they are dealing with a VSA, and prevent discouraging
adoption by avoiding human-like avatars. Start by giving users a choice between humans and
VSAs, then use that experience to study attitudes and adoption.

Virtual support agents without access to a rich source of knowledge cannot provide
intelligent responses, forcing I&O leaders to establish or improve knowledge
management initiatives
Knowledge management is essential for a chatbot or VSA to provide answers to business users, but
the response can only repeat scripted solutions when based on existing data from a static
knowledge base. Looking forward, there is also the potential for VSAs to learn and adapt through
the use of AIOps. Successful VSAs will also leverage data and trends generated by AIOps tools and
techniques to generate new solutions and add to the existing knowledge base. A chatbot or VSA
can learn existing knowledge, but VSAs will only be able to predict and surprise when AIOps is
utilized (see "Applying AIOps Platforms to Broader Datasets Will Create Unique Business Insights").

At their core, virtual support agents are an extension of existing systems presented through a more
human-centric user experience. For task-based actions (such as opening a new incident ticket or
resetting a password), a process ontology is required that maps user responses to fields in
traditional forms. In other words, users are provided a series of natural-language prompts asking for
information that is fed into a form and/or performs an action on the back end. This works well,
because there's a clearly identified and understood context. While there may be substantial
variation in the path taken, the starting point, steps to be taken and a defined outcome are well-

Page 8 of 13 Gartner, Inc. | G00320107

 known and easily described. As such, task-based virtual agents are currently limited to the confines
of the tasks to which they've been programmed.

Unless they are intentionally programmed to accommodate additional tasks (for example,
presenting responses for requesting incident resolution, training or other knowledge-based insight)
directly relevant to the last task, the business user's ability to navigate easily will end up suffering —
and so will their perception of interacting with the VSA. For a more immersive experience, virtual
support agents must also be able to address these broader support questions, without putting the
burden of learning how to navigate on the shoulders of the business user. A rich, extensive set of
instructions is often fine, until someone that isn't an expert (or the original designer) attempts to use
it.

This requires having an established knowledge base to map to so that the VSA can identify and
direct the business user to relevant articles around or directly integrate the content of the articles
into its responses. For IT organizations looking to deploy VSAs, this presents three key challenges:

1. IT organizations lack formalized processes and incentives to maintain relevance of their

knowledge base and as a result their established KM repositories are outdated.
2. Knowledge articles are designed for an IT audience and do not translate well for the business
users.
3. Monolithic articles don't scale. They must be broken into smaller chunks that are relevant to
specific audiences/customer segments, scenarios and supported services — starting from the
most common questions that a business user might ask, down through the lowest level of
technical detail.

Recommendations:

■ Establish a foundation in knowledge management by using techniques like Knowledge-

Centered Service (KCS) before implementing chatbot or virtual support agent technology (see
"Three Steps to Achieve Success With ITSM Knowledge Management").
■ Avoid conversational dead ends by directing users to resolution where possible or automating
escalation to traditional channels where the knowledge responses fail to satisfy the issue.
■ Ensure logic is embedded into the chatbot that collects user feedback to identify relevance of
knowledge responses.
■ Use analytics capabilities of AIOps tools and technologies to enable VSAs to learn and generate
new solutions beyond the static scripting of existing knowledge.

Current implementations of AI on IT service desks are limited to simple chatbot transactions,

so I&O leaders have no fully proven cases of virtual support agents to optimize IT support

Most vendors that sell chatbots and VSAs have limited expertise in IT support and IT service
management or, at best, a superficial level of knowledge. We did not find any products that had
common ITSM processes that support common ITIL workflows beyond incident, request and (in just

Gartner, Inc. | G00320107 Page 9 of 13

 7
one case) problem management. Most customers are using chatbot products for customer
8
engagement or business process automation.

The few IT service desk implementations are limited to very simple transactions, like password
reset, or to searches for articles related to keywords in a knowledge base. These tools are providing
an alternative user interface for functions that are already well-served by self-service utilities and
portals. The conversational platforms within these offerings enable natural language search and
interaction without human intervention, but there's no evidence to indicate that this user experience
is more effective than text-based search; windows, icons, menus and pointing (WIMP) device
interfaces; or touch-user interfaces of more traditional tools.

VSA products that Gartner examined (see Note 1) exhibited bugs and shortcomings that indicate
the technology is still in its infancy and is several years from the mainstream. We did not see
substantial evidence of true machine learning or adaptability. These tools are heavily reliant on
predefined scripts to be well-defined and in place. While we did see products that could create a
problem record from an incident ticket, they could only do so when directly instructed. True AI-
based VSAs should be able to trigger and fulfill that action themselves, but current solutions require
manual intervention when a new scenario is encountered, and require extensive handholding from a
human operator. Technology limitations are part of the cause, but vendors are also wary about
releasing products that are fully autonomous after high-profile news stories of chatbots being
9
subverted by pranksters (see evidence story about Microsoft's Tay AI disaster).

Most VSA initiatives currently in progress are collaborations between managed service providers
and VSA software vendors. The software is not ready for use out of the box, so the solutions are
being developed in pilot schemes run by outsourced service desks. These partnerships are helping
to develop the use cases, scripts and templates for VSA to be relevant and useful. As stand-alone
products and vendors are not ready, early adopters must develop the solution in tandem with the
VSA vendors or managed service partners to have a greater chance of success until the technology
improves. This deployment should be treated as a pilot technology project, rather than a
commercial off-the-shelf (COTS) solution.

Some IT service desks are providing chatbots now, but not at a level sufficient to replace human
operation without severe deterioration to the quality of support. Such cases are instances of I&O
leaders prioritizing cost over service. If the technology behind VSAs improves, we could see IT
service desks supplemented by AI-based support channels, but it will be at least five years (and
probably a lot longer) before they could replace a human-operated IT service desk. Virtual
assistants are taking longer to implement than vendors claim. One client using a product outside of
the IT support use case was promised that its system would start learning for itself after 2,000
scripted answers provided by consultants, but the tool did not generate any answers even after
7,000 scripts had been input. Plan for an increase in effort for the short to medium term. It will be at
least a year before most companies will see any benefit, and much longer (if ever) before that
benefit will transcend what basic self-service Level 0 practices could achieve anyway.

Recommendations:

■ Augment, but do not replace, traditional human-facing channels until the technology matures.

Page 10 of 13 Gartner, Inc. | G00320107

 ■ Deploy current solutions to use as an alternative search user interface to your knowledge base.
■ Early adopters must develop solutions in tandem with vendors and managed service providers.
■ Expect to wait up to a year before a virtual support agent can begin to meet business-user
expectations.
■ Build an ROI for virtual agent support that centers around business-user impact and not simply
call deflection, as adoption and applicable use cases may be limited initially.
■ Wait for mainstream ITSM vendors to deliver VSAs as part of a broader ITSM suite.

Acronym Key and Glossary Terms

VSA Virtual support agent

ITSM IT service management

Gartner Recommended Reading

Some documents may not be available as part of your current Gartner subscription.

"Architecture of Conversational Platforms"

"Four Use Cases for Chatbots in the Enterprise Now"

"Market Guide for Virtual Customer Assistants"

"Three Steps to achieve success With ITSM Knowledge Management"

"Three Simple Ways IT Service Desks Should Handle Incidents and Requests"

"Applying AIOps Platforms to Broader Datasets Will Create Unique Business Insights"

Evidence
1Searches for VSA-related topics and vendors on gartner.com are trending up (363 hits for
"Amelia," 111 for "Kore," 4,378 for AI, 605 for "virtual assistants") as of 31 March 2017.

2In a 2015 Gartner survey of 3,024 consumers across the U.S., U.K. and China, of those who have
used a mobile app on a smartphone in the past three months, 45% of the respondents in the U.S.
and 32% in the U.K. have used a virtual personal assistant (VPA) on their smartphone.

3 Voice search on Google has doubled in one year from 10% to 20% of all searches on Android and
in the iOS app.

Gartner, Inc. | G00320107 Page 11 of 13

 4Gartner has spoken with 23 I&O leader clients about their use of AI on the service desk between
September 2016 and end of March 2017. None of these I&O leaders had approached the business
prior to contacting Gartner.

5 "Uncanny valley," Wikipedia

6 This research was conducted via an online survey from 23 February to 7 March 2017, among
Gartner Research Circle Members — a Gartner-managed panel of IT and business leaders. In total,
153 members completed the survey. Gartner Research Circle IT and IT-business members were
invited to participate. Respondents needed to have knowledge of their organization's usage of IT
self-service or IT service desk. The survey was developed collaboratively by a team of Gartner
analysts, and was reviewed, tested and administered by Gartner's Primary Research team.

7Gartner conducted market research and was briefed by vendors on their chatbot and virtual
assistant technology related to IT support.

8Gartner has spoken with 23 I&O leader clients about their use of AI on the service desk between
September 2016 and end of March 2017. We also spoke with vendors and their reference
customers about their experiences of using chatbots and VSAs for IT support.

9S. Gibbs, "Microsoft's Racist Chatbot Returns With Drug-Smoking Twitter Meltdown," The
Guardian, 30 March 2016.

Note 1 Sample VSA Vendors

Gartner discovered several vendors offering virtual assistants (including chatbots) for an IT service
desk context:

■ BMC Virtual Chat

■ Creative Virtual V-Person
■ IBM Watson Virtual Agent
■ IPsoft Amelia
■ Kore Smart Bot
■ Nuance Nina
■ SunView Software ChangeGear Bot Responders

More on This Topic

This is part of an in-depth collection of research. See the collection:

■ How to Get Smart About Artificial Intelligence for Digital Business: A Gartner Trend Insight
Report

Page 12 of 13 Gartner, Inc. | G00320107

 GARTNER HEADQUARTERS

Corporate Headquarters
56 Top Gallant Road
Stamford, CT 06902-7700
USA
+1 203 964 0096

Regional Headquarters
AUSTRALIA
BRAZIL
JAPAN
UNITED KINGDOM

For a complete list of worldwide locations,

visit http://www.gartner.com/technology/about.jsp

© 2017 Gartner, Inc. and/or its affiliates. All rights reserved. Gartner is a registered trademark of Gartner, Inc. or its affiliates. This
publication may not be reproduced or distributed in any form without Gartner’s prior written permission. If you are authorized to access
this publication, your use of it is subject to the Usage Guidelines for Gartner Services posted on gartner.com. The information contained
in this publication has been obtained from sources believed to be reliable. Gartner disclaims all warranties as to the accuracy,
completeness or adequacy of such information and shall have no liability for errors, omissions or inadequacies in such information. This
publication consists of the opinions of Gartner’s research organization and should not be construed as statements of fact. The opinions
expressed herein are subject to change without notice. Although Gartner research may include a discussion of related legal issues,
Gartner does not provide legal advice or services and its research should not be construed or used as such. Gartner is a public company,
and its shareholders may include firms and funds that have financial interests in entities covered in Gartner research. Gartner’s Board of
Directors may include senior managers of these firms or funds. Gartner research is produced independently by its research organization
without input or influence from these firms, funds or their managers. For further information on the independence and integrity of Gartner
research, see “Guiding Principles on Independence and Objectivity.”

Gartner, Inc. | G00320107 Page 13 of 13