Nameplate / Serial OCR Capture: Snap the Plate, Draft the Asset Record

The problem this kills

A tech is standing in front of a chiller, a pump, or a generator with a greasy metal nameplate, squinting to copy a 17-character serial into a form on their phone. Half the time they fumble a digit, swap an O for a 0, or skip the model number entirely. Back in the office, those typos become duplicate assets, broken warranty lookups, and service history attached to the wrong machine. The "registry" drifts further from reality with every site visit.

The data is right there, stamped on the plate. The problem is the slow, error-prone hop from the plate into your system — and the duplicates and mismatches that hop creates. The fix is to let the tech take a photo, have the app read the plate, match it against the models you actually service and the assets you already have, and draft the record — then put a human in front of it to confirm before anything is written. You do not need to be a developer to build this.

What you'll build

A simple internal web tool for your field-service and asset team. A tech opens it on a phone, photographs the nameplate, and the app runs OCR (reads text from the image) to pull out make, model, and serial. It normalizes the serial (fixes the O-vs-0, dash, and spacing mess), matches the make/model to your catalog and the serial against your existing registry, and shows a drafted asset record with a confidence score on each field. Low-confidence reads are flagged for manual entry, and a duplicate serial throws a warning before you go any further. The tech or controller reviews and confirms the make/model/serial and the catalog match — and only then is the asset created or updated. Confirmed assets export to CSV in the exact columns your system of record expects.

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 — how you capture nameplate data today, the systems and sheets that hold your model catalog and asset registry, what your real make/model/serial formats look like (lengths, prefixes, the characters that get misread), your typical and peak photo volumes, who is allowed to confirm a new asset, and the messy edge cases like worn or stamped-over plates, two plates on one machine, and serials that are genuinely blank. It reads a short spec back to you for a thumbs-up, then builds the tool around your equipment and conventions instead of a generic template. From there it walks the agent through the data model, the photo-capture and OCR extraction step, the catalog match and serial normalization, the duplicate-serial guard, the human confirm gate, the create/update flow, and the CSV 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 asset function needs: login so only your team can use it, row-level security so people only see their own organization's assets, a complete audit trail of every photo, extraction, confirmation, create, and update (who, what, when), a hard human-confirmation gate so no OCR read ever creates or updates an asset until a person signs off, and duplicate guards — a unique serial per make/model plus a duplicate-serial warning at confirm time and a dedupe check on the serial — so the same machine can't be registered twice. The tool exists to make a careful human decision fast: AI reads the plate and proposes the record, a person confirms the call.

Who it's for

Technicians who are sick of thumb-typing serial numbers off a dirty plate, asset controllers who keep cleaning up duplicate records, and service managers who need the registry to actually match the equipment in the field. If you can describe how you name and identify a piece of equipment, you can build this.

You've got this — open the plan, paste the first prompt, and you'll be snapping your first real nameplate into a clean draft record by the end of the weekend.