ServiceNow boasts its AI bot is resolving 90% of its own help desk tickets ServiceNow claims it has created an AI agent that is currently solving 90 percent of the inbound IT tickets to the company's own employee help desk. ServiceNow said, in addition to using the Autonomous Workforce bot internally, select ServiceNow customers are testing it as well. The firm expects to have the product generally …
“How does it know it got the right answer? Because the outcome is measurable inside the same platform, Did the ticket resolve? Did the workflow complete? Did the approval get the right sign-off?
The fact that a ticket is closed doesn't mean it got the right answer. They'll discover that when an attacker gets a password reset or access authorised.
Well, yes.
Problem is, from management and accounting's perspectives, "customer satisfaction" is hard to count. Whereas minutes per call (tickets closed per day, etc.) is easy. The fact that it's a nearly useless metric doesn't matter.
IME once management gets enamored of "metrics" (business school word for "counting stuff", apparently) things are going to start going downhill -- "Managing by the numbers" is apparently favored over knowing the business, understanding your team's mission, talking to people, and so on.
I didn't come up with it, but "Metrics Minded Micro Managers" was a term I heard used.
If A1 were going to replace anyone, seems like it ought to be management types who believe "count stuff" is how to do their job. Computers ought to be able to work that out well enough, and presumably cheaper and round the clock, eh? To say nothing of the Big Wheel Executives who appear to be just making it up as they go along (another form of "hallucination", perhaps?) -- they're even more expensive, including when they're finally shown the door to "spend more time with family".
Any metrics on how many get sent to the wrong people for sign-off, who are very aware of the pressure to make the LLM seem to work and are routing all its mistakes to the right person? Such as the L2 human who should have seen the ticket before it got anywhere near asking for signoff...
bugs and bloatware AI creates are so huge that it difficult to review them all. so it is easier just to use bot to auto-close them by " your problem look like this other one, so it is closed". many gits, specially AI ones, are doing this now. even real issues dont get looked at anymore. it makes no sense to use your time to report issues anymore.
So the people pushing this system thinks it resolves 90% of tickets quickly... I wonder what the users think?
Anyone that has used a chatbot that says 'Is there anything else I can help you with?' when it completely didn't help with the first thing has lived this dream.
The last time I had to deal with one of those "assistants" was when trying to ring Premier Inn (a UK / Europe budget hotel chain) to enquire about a missing booking email. I was sat in an airport and all the background noise confused it immediately, with nothing but a string of "I'm sorry, I didn't understand that, Can you please repeat again?" responses! Thankfully the Artificial Idiot gave up and put me straight through to a person who resolved things within five minutes.
And at what cost? Both org's and the planet have a significant cost as all ServiceNow AI stuff is on a top tier license meaning serious money to access them. Then really how much CO2/water is consumed by LLMs doing basic stuff. Using AI for real value add stuff is debatable but will continue with all the dollars that have been thrown at it, just chucking basic crap through it is never going to let corps with their Net Zero policies get anywhere near if they are honest enough to add the SaaS impact into their overall footprint.
> The internal tickets include high volume issues such as password resets, account unlocks, software access requests, email issues, and VPN connectivity, handled end-to-end within defined permissions and escalation paths.
You don't need AI to solve those, you just need a nifty Perl script.
What I'm not seeing here is how the problem is actually solved/resolved. Does the AI send a sent of instructions to the user? Connect to their machine and physically clear a cache, rename/delete and OST file? I'm guessing that its going to send the users a document of steps. I've seen this before and in some organisations it works but in others it doesn't because the user wants the human touch. But that doesn't matter to the bean counters.
The said part is the thousands of 1st line support staff who are going to lose their jobs and not have enough skill set, or be able to afford the courses to be able to move to the next level.
But company profits matter more than a human life and its been that way for a very long time.
Where I work, we also have a AI chatbot. It solves about 25% of our chats. This is mainly due to stupid questions that can easily be answered before connecting to a real human. In that sense, it is more triage than answering.
Our experiences with the Service Now helpdesk have been such that we tend to give up after hearing that they don't know either, or that they cannot help with that particular problem. If that is the nature of your helpdesk, you might be able to get the 90% score. However, solving a ticket is not the same as solving the customer's problem.
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2026
