OCR
Supplier invoice OCR: every line extracted, every price checked
Supplier invoice OCR (optical character recognition applied to an invoice) reads the text. The Capture agent goes further and works at line level: item reference, quantity, unit, unit price, discount, tax rate, each value returned with its exact position in the page. Extraction is the first step, not the result.
The problem
An invoice read well and checked badly is still an invoice paid in error
The supplier invoice is the most standardised document of the purchasing cycle, and the most treacherous. Every issuer places its columns where it likes, stacks discounts in two places, spreads VAT (value added tax) line by line or at the foot of the page, and slips carriage into an item line. A tool that returns the number, the date and the total leaves everything else to the accounts payable team. We take the problem from the other end, and the overview of purchasing document extraction says it plainly: reading only matters if it makes the check that follows possible.
Because the risk is not in the total. It sits on line twenty-eight of a forty-line invoice, where a unit price runs three cents above the rate signed last year, where a volume discount has quietly disappeared, where the same delivery has been billed twice under two different numbers. Nothing on the page points to it. The document is legible, the arithmetic adds up, and the invoice is wrong. That is why we extract twenty-five fields where the number, the date and the total would be enough to mark a page as “processed”.
What we extract from an invoice
Every field below is expected by default, with no template to build per supplier. They are returned in document order, each value tied to the exact place where it was read, from native PDF as well as scans and photographs.
- Header
- The identification of the document and its link to a commitment: this is what lets you find the matching purchase order, or establish that there is none.
- Lines
- The heart of the check. Each line is extracted separately, including when the table runs onto the next page and the column headers are not repeated.
- Discounts and invoice footer
- The advantages granted outside the lines, which tools focused on totals attach to nothing.
- Taxes
- A single page can carry several rates and several regimes, including reverse charge in construction and intra-community purchases.
- Totals
- Reconciled against the sum of the extracted lines. A total that does not add up is flagged, never adjusted in silence.
- Settlement
- The terms that decide the payment date and the benefit of an early-settlement discount, and the details the money will be sent to.
Twenty-five fields, on a document most tools reduce to five. That granularity is not a stylistic exercise: you cannot verify a unit price you never read, nor a discount you never attached to its line.
The check that follows
Unit price checked against the rate schedule
Each invoiced line is compared with the rate schedule or price list in force on the date of the service, not the one in force today.
Volume discount verified
Is the discount set out in the contract actually applied on this invoice, at the right rate and on the right calculation base?
Tiered pricing checked
Cumulative volume for the period crosses a threshold: the applied price must follow. It is the most frequent discrepancy, and the least visible to the naked eye.
IBAN compared with the supplier master data
An IBAN that changes between two invoices is raised before payment, not after.
Duplicate detected
The same document received through two channels, or rebilled under another number: matching is done on the content of the lines, not on the invoice number alone.
Every discrepancy then opens onto its evidence: the contract clause and the invoice line, side by side, with the calculation shown. Extraction on its own is self-serve and is billed on your monthly volume. The check is scoped with you afterwards, on a quotation.
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 plan covers extraction. Checking your negotiated terms is the next step and is sized with you, because it has to be connected to your contracts, your tolerances and your approval routes.
See pricingNothing 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
Is a forty-line invoice extracted in full?
Yes, line by line, including when the table runs across several pages. The total is then reconciled against the sum of the lines read: if it does not add up, the invoice is raised for review instead of being adjusted in silence.
Does a template have to be built for each supplier?
No. No model to train, no zone to draw: fields are recognised by the role they play in the document, whichever the issuer, from native PDF, scan or photograph.
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.
How do I check a Zylio conclusion?
Every discrepancy opens onto its evidence: the contract clause and the invoice line, highlighted side by side, with the calculation shown.
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
Carry on through the OCR hub
- Delivery note OCRThe document that says what actually arrived, quantity by quantity.
- Purchase order OCRThe reference for what you committed to, line item by line item.
- Supplier statement OCRWhat your supplier believes you owe, as at the statement date.
- Line-by-line invoice extractionWhy a correct total can hide ten incorrect lines.
- Accuracy and confidence scoreCertainty is checked field by field: a score per field, the exact position of the value in the page, a review threshold you set yourself.
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