Agents in the Trades
Dispatch

The 8 a.m. board that assigns itself — and backfills its own cancellations

It's 8 a.m. in a Toronto HVAC and plumbing shop, the phones light up on twelve inbound lines, and the dispatcher has to choose between answering the queue and getting 25 techs on the road. The board she's staring at is already proposing who goes where.

Source: Builts AI, "Home Services Dispatch and Voice AI Agent Case Study" — https://builts.ai/blog/home-services-dispatch-and-voice-ai-agent-case-study/

The problem

Dispatch fails in the first hour. Calls stack up while the dispatcher is still matching jobs to techs, so routine bookings wait, emergencies get whatever slot is left, and some jobs get moved later because the skill or drive time never fit. The board looks full. It is not right.

The second leak is the empty slot. A customer cancels at 9 a.m., and by the time anyone works the waitlist by hand, the nearby tech who could have taken it is already across town. The work exists, the capacity exists — the matching happens too slowly to matter.

How the agent workflow works

  1. Tag the board honestly first. Every tech is tagged by skill tier and zone, and job codes map to a minimum tier — tankless diagnostics, for example, might require a mid-tier tech inside specific zones; warranty work respects brand certifications.
  2. Set a priority ladder. No heat, an active leak, and no hot water get same-day targets; maintenance and warranty checks get next-day targets. Priority decides how wide the slot search runs.
  3. Let the agent propose, not assign. Reading live availability, the agent proposes a tech and slot that respect a load-and-unload buffer and a maximum hop for same-day emergencies. The dispatcher accepts, adjusts, or overrides with one click — and every proposal logs the rule that fired.
  4. Confirm by one tap. The customer gets a text and email to confirm or reschedule, with reminders on a set cadence before the visit — confirming takes seconds, not a phone call.
  5. Backfill the cancellation automatically. An opened slot is offered first to nearby waitlisted jobs, ranked by travel-time fit, membership status, and urgency. The empty hour becomes a booked job before dispatch notices it was empty.

The vendor reports that, over its 60-day window, first-touch bookings rose from 41% to 67% after this rollout — that figure is the vendor's own reported result, attributed here, not a promise for your shop.

What you need to set it up

Where it breaks

Bad tags, bad board. Most early misassignments in the source case traced to vague or missing skill tags. The agent will propose a wrong match confidently, and at scale.

Edge cases need a human door. Out-of-area addresses, attic-only access, and anything that smells like an emergency must pause the automation. The dispatcher who owns exceptions is the job this creates, not the job it removes.

One-tap cuts both ways. Easy rescheduling means more rescheduling, and the backfill only works if your waitlist is current; a stale waitlist turns automatic offers into automatic no-answers.

Overrides ripple. Move one job by hand and downstream jobs need re-checking, or the fix creates the drive-time collision the system was built to avoid.

The bottom line

Run this if your dispatchers spend the morning on routine matching instead of exceptions. Start with one service line, clean the skill tags first, and keep the human override one click away. The win is not a board nobody touches — it is a board where the first proposal is usually right, and a cancellation stops being a lost hour.

Related use cases

← All use cases

One working agent workflow, every day.

Real AI agent use cases from solar, construction, and field operations — what they do, what they cost, and where they break. No hype, no theory.