ManageEngine has published its latest piece of research, The advent of AI agents in ITSM: Perception and future impact (registration required). Of the 300 respondents to the survey, all of whom work for businesses with over £10 million in revenue, 62% said, “AI agents for IT service management will change their organisation’s hiring plan.” It is a stark reminder that AI is here to stay and, importantly, is now affecting hiring practices.
Kumaravel Ramakrishnan, Director of ITSM Marketing at ManageEngine, said, “This research reinforces the fact that the future of IT is human-led and AI-accelerated. AI agents, we believe, aren’t here to replace the workforce -they’re here to transform the workloads, to relieve IT teams of repetitive and manual tasks so they can focus on high-impact work that goes beyond every day operations and includes more strategic initiatives.
“Of course, as organisations gear up to leverage AI technologies, CIOs and CTOs should ensure proper guardrails are devised and put in place to ensure that AI agents operate within the necessary security policies and privacy standards.”
Ramakrishnan’s view is important when compared to data from Orgvue. That revealed that 55% of businesses that made redundancies because of AI, got it wrong.
The survey throws up some interesting facts
Unlike many surveys, ManageEngine kept the survey short with just 10 questions. That explains why they had maximum responses for almost every question. As such, while the number of respondents is small, just 300, the numbers deliver such interesting results. There is also some sensible analysis in the survey rather than just relying on a quantitative approach.
Rate of adoption
82% say they have already implemented AI features and capabilities within their ITSM practice. That’s a very high adoption rate, especially as the UK and other European countries are often said to be substantially behind the US. What ManageEngine goes on to say is that adoption is less prevalent in organisations with over £250 million in revenue.
There is also a difference between industry sectors regarding the use of AI. Finance, healthcare, IT and telecoms are all high adopters. They are all industries where processes are likely to be mature and ripe for automation.
Other sectors, such as architecture and engineering, are in the anti-AI camp. While the survey didn’t go into more details as to why, the likelihood is that this is more about protecting professional status than improving processes.
As a result, of those using AI (246 respondents), 82% were focused on process optimisation, 46% on risk advisory and 42% on knowledge discovery. Risk advisory might be the key to winning over architecture and engineering.
The report says, “virtual agents for end-users and virtual assistants for IT staff – was only in joint fourth place (along with problem prediction) at 39%.”
Feature usage depends on an organisation’s size
This question deserved a qualitative approach to get further understanding into the why. According to the survey, the top three AI features currently in use differed by organisational size as follows:
- 100 – 249 employees – Risk advisory, Knowledge discovery, and Conversational virtual assistant
- 250 – 500 employees – Risk advisory, Process optimisation, and Intelligent categorisation (all joint top)
- More than 500 employees – Process optimisation, Problem prediction, and Knowledge discovery
ManageEngine says there is a caveat here based on the number of respondents from each group. 20% came from those with 100-249 employees, 28% from those with 250-500 and 52% had more than 500 employees. However, it didn’t provide any more details as to how it thought the size altered the capabilities used.
Other surveys at trade shows in the last year have shown that process optimisation is king in most early AI deployments. The last two decades have seen multiple technologies brought to bear to improve process optimisation. AI is just the latest, and indeed promising, tool.
Respondents claimed familiarity with AI Agents
It would be hard for anyone in IT to say they didn’t know what AI Agents were, given the amount of media coverage and announcements from vendors. As a result, 99% claimed to be familiar, with 64% being very familiar.
That vendor push around AI Agents seems to be having the biggest impact on the C-Suite, according to the report. But that might not be good news for those who see AI Agents as the future. While the C-Suite was the most likely to have heard of AI Agents, it was also not the most supportive.
Where are people using AI Agents?
That’s an easy question to answer. The top three uses are:
- Process mining and workflow generation (50%)
- Script generation for process automation and customisation (50%)
- Drafting and documenting post-incident reviews (49%)
However, it is the fourth use – complex workflows like major incident response (44%) that is perhaps far more interesting. There are very few vendors with a real AI Agent solution for incident response. That’s because it crosses so many business areas.
For most organisations, their planning in this area is often incomplete and rarely rehearsed. With so many saying they are using AI Agents for this, it needs a more detailed look at how. Maybe ManageEngine will do this next year.
Using AI Agents for service management comes with challenges
Unsurprisingly, the majority, 93% of organisations, are open to using AI Agents in services management. However, almost all said that there were challenges that had to be overcome before any deployment took place.
The top five concerns were:
- AI governance, data security, and privacy concerns – 45%
- Reliability of AI agents – 39%
- Implementation complexity – 34%
- Unproven technology – 33%
- Unclear ROI – 29%
Some of these are to be expected, such as governance, privacy and reliability. The implementation complexity is interesting, especially given the current state of customer help desk technology.
Extensive integration is already in place between data sets to deliver a single view of the customer to agents. Codifying that for an agent does not seem to be hugely complex. Exactly what the additional complexities are would have been interesting.
The fifth placed concern, unclear ROI, is a surprise. The cost of help desks and building out that technology is expensive. Is this more a comment about the current state of AI Agent technology in customer service rather than ROI? Again, an opportunity for a deeper dive was missed.
One point that does come out from ManageEngine’s own analysis is trust. What permissions should be granted to AI Agents? How would that be controlled? Should there be a human-in-the-loop (HITL) provision? 49% say yes to the need for HITL. Perhaps trust in the machine is far from there.
Are humans a thing of the past?
Interestingly, the survey asks outright, “Do you think AI Agents will replace humans in the IT workforce?” 59% said yes, with 37% saying no. The remainder didn’t know. Given the results of the earlier Orgvue survey, rushing in to make that decision could rebound badly.
Replacing humans is also a tricky issue. If you are talking about process automation and using AI to manage multiple automated processes, that is an easy win. But humans are subject matter experts. They hold a lot about how their job functions in their heads. It is information that is never codified.
Ever since the discussions around Expert Systems in the mid-1980s, the question has always been about how to capture that knowledge.
It is not about the technical knowledge. It is about the use cases, especially the edge cases, which is where the human factor excels. The ability to think outside the box, to make a judgment not based solely on bits and bytes but on other, softer and less tangible factors. It will be interesting to see how this plays out.
For the C-Suite, it’s a done deal. Get rid of the people and sweat the IT asset. For IT directors, it was a different story. They see the HITL as a necessary part of the solution, although how many humans and where they sit is a more complex issue.
Enterprise Times: What does this mean?
This is an interesting survey, as much for what it tells us that we already know as it is for what we perhaps thought but weren’t sure of. It puts some numbers of attitudes of different sizes of businesses, both in terms of employees and revenue.
The ManageEngine analysis also looks at it from the lens of private vs public organisations. Governments are considering using AI to reduce costs, but implementation and ROI will be watched carefully.
From an ITSM perspective, it seems that we are still focused on improving the automation of processes, a task we’ve been at for some time. The much vaunted AI Agent providing help desk cover is a patchy solution at present and has much to overcome. How those challenges are overcome will speak more to the maturity of AI and data governance inside organisations.
When ManageEngine conducts this study next, it needs to allow time to do some qualitative work and not just basic analysis of the numbers. Several areas need further investigation, and each has the ability to shape AI deployment strategies.
