Agents in the Trades
Communication

A photo of the leak books the job

It is 10 p.m. and a homeowner is kneeling in front of a leaking water heater, towel on the floor, sending a photo of it over WhatsApp to the plumbing shop whose number is still on the side of the unit.

Source: Plumbing Perspective, "FieldCamp Launches AI-Powered Field Operations Platform for HVAC, Plumbing and Electrical Contractors" — https://plumbingperspective.com/fieldcamp-launches-ai-powered-field-operations-platform-for-hvac-plumbing-and-electrical-contractors/ — the article describes a water-heater photo sent at night, identified, priced from the service catalogue, matched to a technician, and returned as a confirmed booking, with staff able to step in at any point. Adapted here for a small shop.

The problem

Most after-hours jobs do not start with a tidy form. They start with a photo, a short message, and a homeowner who wants to know if anyone is coming and what it will roughly cost. In a small shop that message sits unseen until morning, while the homeowner keeps calling down the list.

The morning catch-up is not much better. Someone has to work out what the photo shows, find the customer record, check who is qualified and nearby, and only then start the back-and-forth about times. The work is simple, but stacked, manual, and late.

Get one workflow like this, every day.

The free Agents in the Trades newsletter — each workflow broken down: what it does, what it costs, and where it breaks.

How the agent workflow works

  1. The photo is read, not just received. The agent looks at the image and message together, identifies the likely job type — a leaking water heater — and asks only for what the photo cannot show.
  2. The customer’s history comes with it. Past visits, address, equipment, and notes are pulled into the same thread, so a returning homeowner does not retype the basics at 10 p.m.
  3. A price is built from the shop’s own catalogue. The estimate comes from the catalogue the shop maintains, not a number the agent invents, presented as the catalogue allows — flat, ranged, or diagnosis-first.
  4. A real slot is found. The agent checks the actual schedule for a window that fits the job’s duration and urgency, instead of promising “first thing tomorrow” the board cannot honour.
  5. The technician is matched on purpose. Skill, certification, and proximity decide who gets the job — not whoever is first on a list — and the assignment lands on that tech’s board.
  6. The booking is confirmed in the same channel. The homeowner gets the details back on WhatsApp, and staff can step in at any point, with the full history attached.

What you need to set it up

Where it breaks

A photo can mislead. A stain under a water heater can be the tank, a fitting, or condensate from something else. A wrong read books the wrong duration, parts, and sometimes the wrong tech. Low-confidence reads should ask for another angle, not push ahead.

Catalogue pricing is not a diagnosis. The number sent at night is only fair if the shop is clear about what would change it on site. If that boundary is vague, the booking becomes an argument in the driveway.

Automatic booking fills the board automatically, too. A night of plausible bookings can stack the wrong work into the morning, and proximity matching means little if roster data is stale. The source describes a completed booking in under 30 seconds; speed like that still needs caps on job types and same-day load.

The bottom line

Start with one common job type after hours, where the photo is usually clear and the catalogue price is defensible. The win is not that a machine answered — it is that the homeowner woke up to a real appointment, with the right tech assigned.

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.