Callback & Comeback Detector: Catch Repeat Failures Before You Bill the Customer

The problem this kills

A customer calls back three days after your tech "fixed" their rooftop unit. The dispatcher takes it as a fresh request, a new work order gets created, a tech rolls out, and the customer gets a bill for a job that should have been covered under warranty. Now you've got an angry account, a refund to chase, and — worst of all — no record that this was a repeat failure on a job the same tech closed last week. Your first-time-fix rate looks great on paper while your callbacks quietly bleed margin and goodwill.

Callbacks (the same problem coming back at the same site and asset) are one of the most expensive things in field service, and they're almost always detectable. The clue is sitting right there in your recently completed work orders: same address, same equipment, same complaint, closed a few days ago. You just need something that checks every new request against what your crews just finished — before a billable work order is created. You do not need to be a developer to build that something.

What you'll build

A simple internal web tool for your dispatch desk. When a new request comes in, you enter the site, the asset, and the problem. The tool instantly checks it against your recently completed work orders and scores how likely it is to be a callback — a repeat of a job your team just closed. Each flag shows the original job, the tech who did it, the close date, and a plain-English reason. Your service manager reviews the match and decides how to bill it: warranty, no-charge, or new job. Only then is the work order created — tagged with the right classification and linked back to the original job and tech, so first-time-fix and warranty costs are tracked accurately. Everything exports as a clean CSV for your dispatch or ERP system.

What's inside the Implementation Plan

The downloadable plan is a step-by-step file you paste into an AI coding agent. It opens by interviewing you about your business — your intake process, the systems and spreadsheets you use, the exact way your sites and assets are named and numbered, your trades and their typical warranty windows, and your real rules for what counts as a callback. It reads a short spec back to you for a thumbs-up, then builds the tool around your answers instead of a generic template. From there it walks the agent through the data model, importing completed-job history, the new-request intake form, the callback-matching engine (with a configurable callback window per trade), the manager review-and-classify screen, the human approval gate, and the work-order export. Every step ends with a ready-to-copy prompt.

The governance it includes (this is the point)

This isn't a toy. The plan builds in the controls a real service operation needs: login so only your team can use it, row-level security so people only see their own organization's jobs, a complete audit trail of every classification and override (who, what, when, and why), a hard human-approval gate so no work order is created until a manager decides how to bill it, and duplicate guards so the same request can't be processed twice. The whole tool exists to make a careful human decision easy — the AI raises the suspicion, a person makes the call.

Who it's for

Dispatchers, service managers, and quality/warranty leads who own work-order intake and are tired of finding out a "new job" was actually a callback after the invoice already went out. If you can describe what makes two jobs "the same failure" in your world, you can build this.

You've got this — open the plan, paste the first prompt, and you'll be catching your first real callbacks at intake this afternoon.