Supplier delivery notes and invoices load themselves into the system you already have. You confirm, and it stays on record who did
For SMEs with 10 to 50 employees that retype incoming documents every day: lines, quantities, batches, prices. AI reads and proposes, a person confirms, documents never leave the company.
Do you recognise at least two of these?
- Every morning someone in the office retypes yesterday's delivery notes line by line, and the "real" stock is always a day behind.
- Supplier invoices arrive as PDF and XML, and the match with the order is done by eye.
- A batch arrives with the supplier's note and its batch number reaches the system three days later, or never.
- A typo on a purchase price is discovered at month end, on the margin.
- You evaluated a cloud service that reads documents and stopped when you saw that documents leave the company and you pay per page.
If you said yes at least twice, the problem is not the people: it is that the document enters twice, once on paper and once by hand.
What does retyping documents really cost you?
Do the sums on your company, not on the average. Demo estimates, to be verified on your documents.
- A delivery note retyped properly takes 3-5 minutes: with 40 a week that is two to three hours of someone paid to do something else, more than 100 hours a year.
- One wrong line in a hundred: with 2,000 lines a month that is 20 errors a month, each found when it costs most (stock count, margin, complaint).
- A one-day lag between goods received and stock updated is a day in which production and sales work on old numbers.
What is automatic document reading, and how is it different from OCR?
OCR turns an image into text; then a person has to work out which number is the batch and which the quantity. Intelligent Document Processing (IDP) uses models that read text and layout together, such as LayoutLMv3, and return structured fields: supplier, number and date, lines with item, quantity, batch, price. The difference from packaged products is where it lives: inside your system, on your server, with no subscription to another program.
How does it work, in four steps?
- The document arrives by email, from a shared folder or as a photo taken at goods-in with a phone.
- AI reads it and matches each line to items and to the supplier's open order, flagging differences (quantity, price, missing batch).
- A person confirms from the system screen: the document sits next to the proposed fields; correct if needed, confirm. It stays on record who confirmed and what was corrected.
- The system records the stock movement, batch and purchase cost. Traceability and invoice checks start from there.
Corrections improve the model on your documents: suppliers are always the same, and after the first weeks exceptions drop.
Why inside your system, and not in another program?
- Documents stay in the company: the model runs on your server or hosting. Nothing to add to the GDPR processing register, no third party seeing your purchase prices.
- No per-document fee: the cost is the project, not the pages. With 2,000 lines a month the difference shows quickly.
- Human confirmation is in the flow, not in a separate portal: AI proposes, a person confirms, and it stays on record who did. It matters in audits too.
- It starts from a module already in production: document reading is one of 62 ready modules, adapted in about two weeks, then tailored to your suppliers.
What do you need in the company?
A management system to talk to (yours, whatever it is: the module sits next to it, it does not replace it), a server or hosting to run it, and ten real documents for the trial. No GPU, no camera, no dedicated hardware. The accuracy to expect comes from the trial on your documents, not from a brochure: you measure the share of fields right first time, supplier by supplier.
How much does it cost and how long does it take?
First we measure, then we build, and the price comes after the value.
- 30 free minutes on your documents or data: you get a document with today's numbers, yours even if you stop there.
- Clickable mockup in 15-20 days, from 1,500 to 3,000 €, fully deducted from the total if you go ahead.
- Written quote after the mockup, paid in stages with a working demo every two weeks. Your code, on your server, with tests and documentation.
By the day, for teams that want the work inside their own repository: from 850 to 1,100 € a day for applied AI and industrial systems, from 600 to 800 € for a senior Laravel developer, from 700 to 900 € for application security. No per-document fee, no per-user fee. The full price list.
The next step is small: 30 free minutes with ten documents or one month of real data. If it makes sense, a clickable mockup you see before investing. Starts are scheduled one at a time.