The challenge
Every inbound servicing ticket needed a human to read, categorize, and route it before real work could start. At volume, that triage step alone ate hours a day and delayed the cases that mattered most.
What they tried
The team piloted ChatGPT to draft a first-pass summary and suggested category for each ticket — a starting point an agent could confirm or correct in seconds instead of composing from scratch.
The result
Triage time dropped by roughly half, and agents reported spending far more of their day on genuinely complex cases rather than sorting the queue.