Applied machine learning that routes and drafts responses for inbound support tickets, with a human always in the loop.
SaaS engagement, 10 weeks in scope.
Support volume was outpacing headcount, and generic chatbot deflection was hurting satisfaction scores.
Built a triage model trained on the client's own resolved-ticket history, routing tickets to the right specialist and drafting a first-pass reply for review rather than replacing the agent.