AI Support Assistant
An LLM-driven IT support assistant at my current healthcare employer. It works in Google Chat, an internal dashboard and Jira, resolves about 80% of tickets on its own, and hands the rest to a person by design.
The Challenge
I run IT for a 75-person healthcare company as a one-person function: identity, devices, SaaS, security and compliance. Every support request lands with the same person who owns the audits, the vendor contracts and the roadmap.
The Approach
I built an LLM-driven support assistant and put it where people already ask for help: Google Chat, an internal dashboard and Jira. It resolves the requests it can on its own. The ones it can't go to a person, on purpose. The handoff is part of the design, not a failure mode.
The Outcome
The assistant resolves about 80% of tickets on its own. The rest reach a human, because support at a healthcare company still needs a human in the loop.
Key Details
- Works in three places: Google Chat, an internal dashboard and Jira
- Resolves about 80% of tickets without a person
- Hands the rest to a person by design, so a human stays in the loop
- Built and run inside a one-person IT function serving all 75 employees