OCR
Multilingual and multi-currency OCR: formats stop blocking anything
Multilingual invoice OCR (optical character recognition) does not stop at characters: a date, a thousands separator, a currency symbol and a tax mention all have to be interpreted. The Capture agent normalises everything on output and keeps the original value beside the normalised one.
The problem
The same amount, written six ways
1.234,56 and 1,234.56 mean the same amount, or two different amounts, depending on the issuing country. A date written 03/04 changes month across a border. A currency symbol placed before or after the number does not carry the same accounting weight. Each of these details is enough to falsify a price check, and the role of OCR in the control chain explains where that check happens.
Then come the tax mentions: intra-community VAT, reverse charge, exemption, the customer VAT number, the incoterm. They are not decorative: they decide the amount actually due and the accounting entry expected. With a supplier established abroad, the missing field is almost always one of those.
The subject is not cosmetic: it decides what is comparable. The same reference bought in Spain and in Germany only compares if units, currencies and periods have been brought back to the same basis. Without that normalisation, spend analysis by product family stops at the national border, and grouped negotiation with an international supplier stays out of reach for lack of defensible figures. The gap is rarely one of goodwill; it is one of formats nobody had time to reconcile.
The result, in most organisations, is a two-speed process: domestic suppliers go through automatically, the others go to manual entry. The controlled scope therefore stops at the border, which is often exactly where the negotiated terms are most complex.
What is normalised on output
The value read and the value you can work with are two different things. Both are kept: the original as evidence, the normalised one for the calculation.
- Language of the document
- Column labels, item descriptions and invoice footer mentions are interpreted in the language of the document, with no per-supplier template.
- Currency and symbol
- The currency is identified per line and per total, from the symbol as much as from the three-letter code. No implicit conversion: the amount stays in its original currency.
- Exchange rate
- Where a comparison requires it, the rate applied is explicit, with its reference date. The converted amount never overwrites the invoiced amount.
- Date formats
- Day-month-year, month-day-year, year-month-day, written-out dates, week numbers: reduced to a certain date, or flagged as ambiguous.
- Thousands and decimal separators
- Full stop, comma, thin space, apostrophe: the number is rebuilt according to the conventions of the document, not those of the reader.
- VAT mentions and applicable regimes
- Intra-community VAT, reverse charge, exemption, the VAT numbers of both parties: extracted as fields, never left as free text.
- Units of measure and packaging
- Metric or imperial units, local packaging: brought back to the unit in your reference data, with the conversion factor displayed.
One case deserves detail, because it recurs constantly: the invoice issued in one currency and payable in another. The document then carries two amounts, a rate and sometimes a conversion date. All three are extracted separately, and the check applies to the amount contractually due, not to the one used for settlement. The same logic governs a discount expressed in a currency other than that of the lines, and a deposit paid in a third one.
An ambiguous value is never settled in silence. It is flagged as ambiguous, with the area of the document where it was read.
What it changes
The controlled scope stops ending at the border.
- Foreign suppliers enter the same flow as the others: one process, one queue of exceptions to handle.
- Spend compares from one country to another because units and currencies are explicit, not guessed at analysis time.
- VAT regimes extracted as fields make it possible to check the expected entry before approval, rather than at close.
- The original value stays displayed beside the normalised one: the conversation with the supplier is about their own document.
None of this requires standardising your suppliers. The document stays as it was issued, in its language and in its currency; it is the output that is put into the format of your reference data. Your ERP (enterprise resource planning system) remains the source of truth and receives comparable values, with no development added to it. Nothing is asked of the supplier either: no portal to adopt, no template to follow, no field to fill in on your behalf.
Capture
Gather and consolidate all your procurement data.
Collection by dedicated mailbox, upload, SFTP, scan, API or ERP export. Reads every format (native PDF, scan, photo, spreadsheet, structured feed), then extracts line by line: supplier, references, quantities, unit prices, discounts, taxes, terms.
The end of re-entry. A procurement history you can finally query.
What Capture detects
- Long documents and multi-page tables
- Unknown layouts
- Several languages and currencies
- Poor scans and handwriting
The subscription covers extraction, in every language and currency encountered. Checking your negotiated terms is the next step and is sized with you: the monthly cost for your volume is shown with no commitment condition.
Nothing is lost. Everything can be checked, everything can be proven.
- The contract clause and the invoice line, highlighted side by side.
- Every extracted value stays linked to the exact place in the document where it was read.
- The same case produces the same decision, today as in six months: the rules are applied deterministically.
- No discrepancy is set aside in silence. Anything that matches no rule is raised, with its reason.
- The agent records what it did, in the order it did it: who, what, how much, when.
Frequently asked questions
Does the language of a document have to be declared in advance?
No. The language is recognised on reading, document by document, including when a single batch mixes several. No template and no per-supplier configuration is required.
Are amounts converted into the company currency?
The invoiced amount stays in its original currency. A converted value can be added for comparison, with the rate used and its reference date displayed alongside.
Are reverse charge mentions extracted as fields?
Yes. The applicable regime, the mention carried on the document and the VAT numbers of both parties are extracted separately. An invoice announcing reverse charge without carrying the expected numbers is flagged as incomplete, with the area of the document concerned.
Which documents does Zylio handle?
Invoices, contracts and amendments, price lists and rate schedules, quotes, purchase orders, goods receipts, credit notes, statements, as native PDF, scan, photo or structured format, in several languages and currencies.
Measurable impact in every environment
More than 5 million procurement documents analysed
Between 1 and 7% of margin recovered
on the scope analysed
From 15 to 45% of time given back to teams, per FTE
depending on the scope and on data maturity
Zylio fits into your existing ecosystem.
The ERP runs the process. Zylio handles the exception and recovers the value that escapes it: invoices without a purchase order, line-by-line price discrepancies, duplicates and overbilling, off-contract spend.
Your data under high security.
Zylio meets the most demanding standards, and nothing is committed without your approval.
- Certifications
- Hosting
- Encryption
- Access
Continue on OCR
- Supplier quote OCRPutting offers on a single grid before comparing them.
- Supplier statement OCRWhat is missing and what was settled twice, over a period.
- Multi-channel intakeEmail, scanned paper, portal, EDI: a single intake flow.
- Supplier invoice OCREvery line extracted, every price checked: extraction is the first step, not the result.
See what this looks like on your own data
Twenty minutes, on a spend category of your choosing. We show you what the agents detect, with the evidence behind it.
- No commitment, on your own data
- Result in 3 weeks
- 20 minutes, no sales pitch
- Your data stays hosted in France
- No change of tool or process