The full story
The challenge
A three-person IT team supporting over two hundred staff was always going to be stretched. That is not unusual. What made it worse was the nature of the requests coming in.
At Hartley Group, the IT helpdesk handled everything from business-critical infrastructure issues to password reset requests. Both arrived through the same channel. Both competed for the same attention.
The team were technically capable and conscientious. They were also spending the majority of their day on requests that followed a completely predictable pattern.
Password resets. Software access requests. VPN connection problems. Printer issues. Questions about which application to use for a given task. Each had a standard resolution. Each still required someone to read the ticket, identify the issue, respond with the answer or action, and close the request.
During busy periods, first response times stretched to several hours. Staff waiting on a password reset could not work. The IT team could not triage properly because the volume made prioritisation difficult. Complex issues that genuinely needed technical attention sat behind a queue of requests that a well-written guide could have resolved.
The team understood the problem clearly. They did not have the capacity to solve it while also doing their jobs.
The solution
We built an internal AI bot integrated directly into the company’s existing helpdesk platform. Staff raise tickets the same way they always have. The difference is what happens next.
When a ticket arrives, the bot reads the content and classifies the request. For known issue types with documented resolutions, it responds directly with the correct information or triggers the appropriate action. A password reset request initiates the reset automatically and confirms back to the user. A software access request routes to the correct approval workflow without an IT team member reading and forwarding it manually.
For issues the bot cannot resolve with confidence, it classifies the ticket by type and urgency and routes it to the right team member with the relevant context already attached. The IT team does not start from a blank ticket. They start from a classified, contextualised request.
The bot draws from the helpdesk knowledge base and can be updated as new patterns emerge. When the team identifies a recurring issue type, they add the resolution and the bot handles it from that point forward.
We built the classification logic carefully. A bot that misroutes or misresolves tickets creates more work, not less. We ran the system in observation mode for three weeks, comparing bot classifications against how the team would have handled the same tickets, before allowing automated responses to go live.
Transparency was built in throughout. Staff always see when the bot has responded and can escalate straight away if the answer is not right. The IT team can view every interaction and override anything. The system is an assistant, not a gatekeeper.
The result
Within the first month, 65% of incoming tickets were handled automatically without IT team involvement. Password resets that previously required a human in the loop resolved in under two minutes. Software access requests were routed correctly and approved through the right workflow without the IT team acting as a relay.
The team’s available capacity shifted considerably. The hours previously spent on repetitive responses went to infrastructure work, security improvements, and the complex issues that actually required their expertise.
First response time improved for all tickets, including those needing manual handling, because the queue was shorter and the remaining items were prioritised clearly.
The feedback from the wider business was simple. Things that used to take hours now took minutes. The IT team were easier to reach for real problems because they were no longer buried in routine ones.
If your support team is spending professional time on predictable, repeatable queries, an AI-powered workflow can handle the pattern and free them for the work that needs them.
What changed
After go-live, the shift showed up in the week first. Then it showed up in the numbers. Here is what that looked like on the ground.
without any IT team involvement
for common requests, down from several hours
instead of routine admin
How we approached this