IT support is getting a new layer of automation that doesn't wait for a human to file a ticket. A company called Serval has released an AI agent, named Catalyst, that scans IT support history and quietly builds automated fixes for problems before employees even report them.
What happened
Catalyst is now generally available and turned on by default for customers of Serval's IT service management platform. Unlike a typical chatbot that answers one question at a time, Catalyst works more like a supervisor. It reviews past support tickets, internal documentation, and even plain-language instructions from IT staff to spot patterns — say, the same password reset issue happening every Monday morning across a department.
Once it identifies a recurring problem, Catalyst doesn't just flag it. It drafts the actual automation needed to fix that category of issue going forward, then deploys smaller AI agents to carry out the fix in the background. The idea is that routine IT headaches get resolved on a rolling basis, without a human technician manually building each solution.
This sits on top of Serval's broader platform, which already uses AI to handle IT service requests. Catalyst is described as a step above that layer — an agent that manages other agents, deciding what's worth automating and then doing the engineering work itself. That's a meaningful shift from AI tools that assist a technician to AI tools that replace parts of the technician's job entirely.
The company is positioning this as a way to shrink the backlog of tickets that pile up in any IT department, particularly the repetitive, low-complexity ones that eat up staff time without requiring much judgment.
Why it matters
Most business AI tools released this year have been assistants — they draft an email, summarize a document, answer a question. Catalyst represents a different category: an agent that identifies its own work and builds the tooling to do it, with minimal human direction. That's the direction a lot of enterprise AI is heading, and it raises the stakes on questions of oversight and accuracy, since the agent is now authoring automations that touch live systems, not just suggesting text.
It also reflects growing pressure on IT departments, which are often understaffed relative to the volume of support requests they get, especially at smaller companies without dedicated help-desk teams.
What this means for small businesses
For small businesses without a full IT staff, tools like this are appealing precisely because they promise to catch problems before they become a fire drill. If your business relies on a handful of overworked people to handle tech support alongside their regular jobs, an agent that quietly resolves recurring glitches could free up real hours.
But there's a practical catch: this only works if you already have decent records of past support issues and some documented procedures for the AI to learn from. A business with messy, undocumented IT history may not see much benefit right away, since the agent needs patterns to detect in the first place.
Cost and vendor lock-in are also worth watching. Tools like this tend to be priced for mid-size and larger IT operations, and adopting a platform where an AI agent is quietly building automations into your systems means trusting that vendor with a fair amount of access and judgment. Ask what happens when Catalyst gets it wrong, and how easy it is to review or reverse an automation it builds.
What to watch
The real test will be how Catalyst performs outside curated demos — whether its automations hold up across messy, real-world IT environments, and whether businesses feel comfortable letting an AI agent make changes to their systems without a person signing off first.
The bottom line
Self-directed IT automation could genuinely reduce busywork for stretched small business teams, but it's worth waiting for real-world reviews before handing an AI agent the keys to your systems.