When your engineers become tier-1

Every fast-growing SaaS company hits the same wall: the people who built the product end up answering questions about it. This one was no different. Tier-1 was drowning in how-to and billing tickets, and the “AI” chatbot they’d bolted on the year before had exactly one reliable move — “Let me connect you to a human.” On paper it deflected plenty. In practice it resolved almost nothing, and the humans it connected people to were, more and more often, engineers.

The first change wasn’t technical. The team retired their deflection dashboard and started measuring the only number that mattered: did the customer’s problem actually get solved? It was a humbling week. The “deflection rate” they’d been proudly reporting turned out, on inspection, to be mostly abandonment — customers giving up, not getting helped.

With the right metric in place, the fix became obvious. They didn’t need a better auto-reply. They needed something that could read the docs, check the account, and actually do the small things — add a seat, fix an invoice, walk someone through setup — and escalate only what genuinely needed a person.

Tier-1 volume is resolved now, not contained. Escalations to engineering fell by roughly two-fifths. The documentation quietly got better, because the gaps showed up in the analytics as the questions the agent couldn’t answer yet. And — the line the VP keeps repeating — the engineers got their Fridays back.