Handwritten job sheets turned into work orders

Phone photos of handwritten job sheets read into work orders, with hours, parts and signatures checked before billing. For HVAC and trade shops in NE Arkansas.

Demonstration using synthetic data

Every record, name, address, and log entry on this page is invented for the demonstration. No client, client system, or production record is involved, and no business result is claimed.

Author
Tanner Rusher
Published
Written for
Trades and service businesses
Example · fictional dataAbout 40 seconds, with music and no narration.

Who this is for

Heating and air, plumbing, electrical and other trade shops in Northeast Arkansas whose techs still fill out paper job sheets in the truck. If someone in the office spends the end of each week reading handwriting and keying hours, parts and serial numbers into work orders before anything can be billed, this is that job.

The problem, in your words

The sheets come back folded, smudged, and sometimes with a coffee ring on them. The tech wrote the hours one way and the clock times another. The parts do not add up to the total. A serial number has an S that might be a 5. Nobody catches it until the invoice is wrong or the warranty claim bounces.

What happens, step by step

  1. The tech snaps a photo of the finished sheet with a phone, at the job or back at the shop.
  2. An AI model reads every field on the sheet and marks how sure it is of each one.
  3. Plain rules check the sheet: the hours match the clock times, the parts add up to the written total, and there is a signature and a phone number.
  4. Fields marked unsure, and any sheet that fails a check, go to a person with the photo beside them.
  5. Sheets that pass become work orders, ready for billing.

What you approve

A sheet with a doubtful field or a failed check does not become a work order until someone in the office has looked at it. The person sees the photo and the reading side by side, so the check takes seconds instead of a phone call to the tech.

What the example run showed

The video is a real run on fictional data: fifteen made-up job sheets from a made-up heating and air company, photographed to look like phone pictures, tilted and shadowed, with crossed-out corrections. 252 of 258 fields were read right. All six misreads were serial numbers the run had already marked as unsure, and the rules caught all four problems planted in the sheets. Twenty-one fields were marked unsure and sent to a person. Those numbers come from one run on invented sheets.

Setting it up in your shop

This is business automation that keeps your techs on paper if they like paper, and takes the typing out of the office. It starts with a 20-minute call about how sheets come back today and where the work orders live. If it fits, a Workflow Review maps the job, and an Automation Pilot builds it around your own sheet. I’m based in Jonesboro and happy to see the sheets in person.

Whether the workflow, the systems, or the approval rules look different in your organization, the next step is the same conversation.