Parseur

You name a field and write a sentence about it, for example "the total amount including tax, near the bottom right". The AI then finds that value on every new document, even when the supplier changes the layout. This is the part that separates Parseur from older template-only tools, and it saves the most time on messy real-world files.

When a document never changes shape, a fixed template with anchored regions is faster and cheaper to run than AI guessing. You draw or select the area, map it to a column, and the parser reads the same spot every time. Most accounts end up with a hybrid parser, with templates for the stable fields and AI for the rest.

Parseur runs OCR on PDFs that contain images, scanned contracts, and phone photos of receipts. Text comes out searchable and usable as field input. Accuracy depends on the scan quality, so a 200 dpi scan of a printed invoice works far better than a phone shot under office lighting.

Raw text rarely arrives in the format your database wants. Post-processing lets you convert date formats, strip currency symbols, multiply quantity by unit price, look up a value in a table, and validate that a total matches its line items. Rules run after extraction, so a document that fails validation can be flagged instead of silently stored.

Each mailbox is its own email address with its own parsers and rules, and you can run unlimited mailboxes even on the free tier. Every processed document is logged with its status, extracted values, and source file, which makes it possible to answer "where did this number come from" months later.

Data leaves Parseur through Google Sheets, Zapier, Make, Slack, Salesforce, HubSpot, Airtable, QuickBooks, Xero, or a plain webhook. Developers who want to skip the UI can use the REST document parser API to send files and receive JSON. That makes automated data entry possible inside an existing back office system rather than beside it.