EA Standard: invoice workflow development update

From invoice data to a usable workflow

A large part of the recent Standard work is happening before anything reaches a polished screen. We are building the mechanisms that create, store, and update invoice drafts, validate their data, and move that information safely between the different stages of the process.

That matters because an invoice workflow is not simply “take a photo and read the text”. A document can be incomplete, ambiguous, formatted in an unexpected way, or contain values that need to be interpreted before they can safely affect products, prices, quantities, or stock.

The goal is a workflow where document capture, OCR, validation, and StockMate’s inventory model behave like one process, rather than a collection of disconnected features.

OCR is only the first step

OCR can identify text and values from a photo or scan, but recognition alone does not make the result trustworthy. The extracted information still needs structure: which value is the invoice number, which date matters, which rows describe products, which quantities belong to which units, and which fields are uncertain or missing.

That is why the current work also covers draft persistence and validation. The system needs to preserve what has already been recognised, allow corrections without losing context, and identify data that should be reviewed before a document is accepted.

Why build below the interface first?

Because a convincing OCR screen is easy to demonstrate and surprisingly easy to get wrong underneath. If the data model, validation rules, and persistence layer are vague, every later screen ends up compensating for them in its own way.

StockMate is taking the opposite route. The contracts and storage rules are being established first, then the visible workflow can rely on them instead of inventing new behaviour at every step. It is slower to photograph for a progress post, but considerably nicer when software is expected to remember what happened yesterday.

What comes next

The next stages will continue connecting these foundations into the complete Standard invoice flow: stronger document validation, wider OCR integration, review and correction of recognised data, and the transition from an approved draft into actual inventory operations.

We are also continuing the surrounding Standard work so invoice processing can fit into the same local-first model as the rest of StockMate rather than becoming a separate subsystem bolted onto it later.