AI-powered functions, like automated ticket tagging, predictive maintenance, and virtual service agents, can cut out manual work and help staff respond more quickly to issues. It can also create a problem. When AI capabilities are introduced before an IT service management (ITSM) environment is ready to support those abilities, companies ultimately orchestrate processes that are flawed, inconsistently applied, poorly documented, or based on bad data. It doesn’t necessarily make IT smarter. Sometimes, it’s just faster at previously flawed execution.
Why ITSM Foundations Matter
AI is only as good as the data and processes it has access to. In a typical ITSM environment, AI would have access to incident tickets, configuration items, knowledge base articles, service catalog information, and historical incident resolution patterns.
If these data points are incorrect or incomplete, AI recommendations will also be incorrect. It is possible for an AI engine to route an incident incorrectly, suggest an outdated resolution, or identify an issue occurring without capturing its business impact.
This makes ITSM maturity one of the key components of any AI strategy.
However, before organizations can deploy high levels of automation, they need to determine whether their core processes are consistent enough to support it. Relevant questions here include:
- Are incidents classified in a consistent fashion?
- Is the service catalog kept up to date?
- Are knowledge articles regularly maintained?
- Is configuration data accurate?
- Can automated decisions be tracked and assessed?
These may sound like fairly standard IT management issues; however, their consequences take on greater significance when AI is in the mix.
The Hidden Risks of AI-Driven ITSM
One of the trickier problems is that AI can make a company’s operational shortcomings less painful to the eye rather than removing them.
An automation-based help desk can handle a large volume of simple requests without human supervision. That seems like productivity. But if many customers never get correct or complete information and the desk sends complex calls to the wrong team for a solution, the employer is just dumping more work further downstream.
There is a governance aspect to consider as well. AI solutions can influence decisions over incident priority, access requests, and resource placement, etc. If left unmanaged or unmonitored, then employees may be left wondering why a particular output or answer was given.
So, organizations looking at AI risk in ITSM will need to consider more than just model accuracy but the quality of data ingested, the processes modeled, accountability, and human oversight.
Building a Safer Path to Automation
A way to think about it pragmatically addresses AI adoption as an evolution of ITSM rather than a replacement for it.
Right now, that means organizations should figure out ways to better the processes that AI support will eventually seek to improve. They should demand that incident categories and standard definitions be established, inactivate those knowledge articles that are outdated, institute better configuration management, and more clearly assign ultimate stewardship over the service.
Then and only then are AI initiatives appropriately brought online.
A helpful sequence is:
- Clean the data. Identify where it’s wrong, where it’s the same, where it’s old.
- Standardize the process. Reduce the extent to which common IT processes have so much variation.
- Find narrow use cases. Test the AI there, make sure it’s not incentivized to make so many mistakes that they’re difficult to monitor or correct.
- Monitor outcomes, not just volume. Can you track correction quality, the rate AI is overruled, or how happy users are that you resolved the issue?
- Allow humans to have oversight. Define situations where employees must review or override automated decisions.
It also makes it easier to figure out which areas AI is a force multiplier vs. where we should put traditional process improvement lenses against, too.
AI and the Next Stage of IT Operations
The real power of AI in ITSM in the long run is unlikely to be just the ability to automate as many tasks as possible. Its better value may come in helping IT teams see patterns, anticipate issues, and make more informed operational decisions.
That, however, takes some planning around where AI fits into an operational model. The technology, data, process, and people have to advance together.
The point is, in many ways, a simple one: AI can make ITSM easier, but it can’t make up for the difficulty of an enterprise foundation in perpetuity. Enterprises that inject automation into strong service management processes will likely be better able to use AI without the creation of multiple new operational risks through their innovation.
Photo By: Saj Shafique; Unsplash
Marcus Whitfield writes about developer tools, programming languages, and the software trends shaping how engineers build. Before joining DevX, he spent five years as a full-stack developer and two more running a small dev-tools newsletter that topped 10,000 subscribers.
