Gateway — document reading
Accounting OCR: from reading (Google Cloud Vision) to an organised file
Yes, Cortex reads invoices, QR-bills and receipts. The reading engine is Google Cloud Vision — a single path, not a “home-grown OCR”. That reading is not Organized Accounting: it feeds structured documents, proposed entries, a review, then export to the software you already use in Switzerland.
Yes, Cortex reads the documents
If you are looking for accounting OCR in Switzerland, the direct answer is: you drop the unsorted batch, Google Cloud Vision extracts the text, then the file continues. OCR answers “does it read?”. It does not answer “is it right, isolated, validated and exportable?”.
OCR is a component, not the product
A reading API returns fields. An accounting file needs identifiable documents, proposed entries, numeric controls, human validation, and export to Crésus, WinBiz, Abacus or another format you already run.
Cortex is not an OCR product. The category is Organized Accounting: OCR belongs there as reading, not as the promise.
What follows the reading
Reading → structured documents → proposed and checked entries → human review → export. Stop at OCR and you still have no file.
What OCR does not guarantee
A blurry scan, an evening photo, a truncated QR: the case stays for review. Cortex publishes no recognition rate. A single figure would be a guarantee that reading cannot keep.
Missing information stays missing. An incomplete line is not stored as an entry.
FAQ
Which OCR engine do you use?
Google Cloud Vision, a single path. It is not a home-grown OCR, and it is not the product.
Is OCR enough to book the entry?
No. You need structured documents, checked entries, your validation, then export.
What if the scan is poor?
The case is flagged as uncertain. You review. Cortex does not invent the missing field.
Does Cortex replace an archive system?
No. Reading is not filing, and filing is not ten-year legal retention. Long-term filing stays yours.