Did you change the model?

No. Same model, same tier, same vendor. Every number in this case came from the original model choice.

Is this a one-time latency benchmark or a sustained result?

Sustained. We watched the class of tail failures across two weeks of post-engagement production traces. The behaviour did not return. The engagement also installed the observability that tells the client when it does return.

Does this work for non-dispatcher copilots?

Yes, when the product is a chat agent walking a user through a structured operational workflow — tickets, rosters, routes, form-fill on structured records. The farther the product is from that shape (e.g. purely open-ended Q&A), the less the pattern applies.

Why is "variance collapsed" more important than "average faster"?

Because operators remember the worst trace of their week, not the average one. A copilot whose average is forty seconds but whose tail is seventy-five seconds generates more support tickets, more "is this thing broken?" messages, and more churn than one whose average is forty-five but whose tail is forty-seven. Collapsing variance collapses the user’s perception of reliability in a way that average speed doesn’t.

Ускорили dispatch-копилот и сделали его в 4 раза предсказуемее — та же модель

body{background:#050510;color:#fff;font-family:system-ui,sans-serif;padding:2rem;max-width:720px;margin:0 auto;line-height:1.6}Ускорили dispatch-копилот и сделали его в 4 раза предсказуемее — та же модельСредний случай стал быстрее. Хвостовые задержки исчезли. Теперь каждый прогон укладывается в узкий коридор — больше никаких непредсказуемых скачков на 12–14 ходов.Continue