Applied AI
InsightGen
From ticket to routing decision
A hackathon prototype for teams processing support tickets.
Project story
During the hackathon, my teammates gathered support-ticket requirements and I built the prototype. One question shaped the design: if two different tickets arrive, how do we choose the right knowledge and model for each? I brought retrieval and model routing together to explore how that choice affects quality and cost.
Screenshot gallery
The problem
Before resolving a ticket, support staff spend time understanding the request, finding relevant knowledge and choosing a handling path. Similar cases can trigger the same work repeatedly.
Measuring the problem
Sample tickets by type and measure hands-on triage and lookup minutes. Multiply by monthly volume for each type to estimate the team’s workload.
The opportunity cost
An hour spent triaging recurring tickets could go toward a complex case in the queue. Recovered capacity depends on routing quality and the human review still required.
How the product helps
The InsightGen prototype retrieves relevant knowledge and routes requests between models. The hackathon explored the trade-off between processing quality and cost.
My contribution
I built the technical prototype while the team gathered business requirements. I combined knowledge retrieval with model routing to explore processing quality and cost.

