Sample: AI Phone Agent Handling a Plumber Lead

This worked example walks through a representative after-hours call to the AI phone agent for a private plumber. It shows the call the agent received, the structured output captured during the conversation, and the reviewer notes that turned a raw transcript into a job lead the plumber could act on the next morning.

The call that came in

A residential customer rang at 9:40pm reporting a slow leak under the kitchen sink. They wanted a same-day callout, gave a callback number, and read out a postcode and a partial address. The line was a little noisy, and one postcode digit was hard to make out.

The agent did not try to book the job or quote a price. Its job on this call was narrow: greet the caller, confirm this was a plumbing issue inside the service area, capture the details that matter for a callback, and set the urgency. That scope is set during after-hours call answering intake, so the agent stays inside it.

What was captured

During the call the agent filled a consistent fields block:

Alongside the fields, the agent wrote a two-line plain-language summary so the plumber could grasp the job without reading the whole transcript.

What the reviewer changed

Before the lead was delivered, a human reviewer checked it against the recording. Two corrections came out of that pass:

That review step is the difference between a transcript and a usable lead. The agent captures fast; the reviewer makes sure nothing wrong or invented reaches the plumber.

The deliverable

The plumber opened one short lead the next morning: a clear issue, a confirmed callback number, a same-day urgency tag, and one explicit thing to confirm on the call. No listening back to a recording, no guessing at a postcode.

This example is intentionally short. It shows the shape of the deliverable, not every call type a real service handles. Volume, hours of coverage, and how deeply each call is reviewed are settled at intake.