OCR
Contract and rate schedule OCR: turning a signed PDF into a checkable price grid
Supplier contract OCR applies optical character recognition to a signed agreement, then goes considerably further: the Capture agent extracts the unit price schedule (the bordereau de prix unitaires, or BPU) line item by line item, its volume tiers, the index clause and its reference index. A signed PDF becomes a checkable price grid.
The problem
The contract says one thing, the invoice says another, and the contract is a PDF
A signed framework contract is a paradoxical object: it is the only document that states what you are supposed to pay, and the only one nobody opens. It sleeps in a storage space, as a PDF, forty pages plus two amendments, with the price schedule in appendix 3 and the index clause in article 9.2. Nobody has time to compare an invoice line with a table of eighty line items. That is exactly what changes with our approach to document extraction: the document becomes data, and data can be checked.
Because a rate schedule is not a price list. It is a price contract: each line item is described with its unit, its packaging and often a condition of application: above a given volume, before a given date, outside a geographical zone. To that are added the year-end rebate (remise de fin d’année, RFA), calculated on a cumulative base that is almost always debatable, and the revision clause, which indexes prices to a published index at a set frequency, sometimes with a cap. Reading a rate schedule is therefore not extracting a table: it is reconstructing the rules that decide the price of a line on a given date.
What we extract from a contract and its rate schedule
The fields below are expected by default, amendments included, on a scanned contract as much as on an eighty-page native PDF. Each value stays tied to the article and the page where it was read.
- Parties and term
- The legal scope: which group entities genuinely benefit from the terms, and by what date a termination notice has to be sent to be admissible.
- Scope and price schedule
- The heart of the rate schedule. A line item without its unit and packaging is not a price: it is a number, and it compares to nothing.
- Year-end rebate
- The year-end rebate (RFA) is calculated on a cumulative base (turnover counted, categories included, entities taken in) that is almost always debatable.
- Price revision
- The clause that moves prices without anybody renegotiating. This is where the most durable discrepancies sit, because they repeat at every revision date.
- Commercial terms
- What decides the real cost of a small order: the free-carriage threshold, the minimum, and the charges the contract allows to be billed on top.
- Commitments and settlement
- What you can hold the supplier to, and what the supplier can hold you to: lead times, service levels, payment terms.
How a revision clause is read
An index clause names an index, a formula and a revision date. The most frequent are the Syntec index for professional services, ICHT-IME for engineering and maintenance labour, material indices for steel, copper or plastics, and fuel indices for transport. Each has its own publication frequency, its base value and its rounding convention.
So we extract the named index, the formula exactly as written, the base value, the frequency and the cap where one exists. That is what makes it possible to establish, later, that a revision was applied to the right index but three months too early: the most widespread discrepancy of all, and the hardest to spot with the naked eye.
Twenty-four fields, six of which decide on their own the price of a line on a given date. A contract reduced to its parties, its term and its headline value is not usable: what gets checked is the pricing rule, line item by line item.
The check that follows
Every invoiced line checked against the schedule
The contract price for the line item, on the date of the service, against the price actually billed. A line item missing from the schedule is raised as such, never force-matched.
Revision verified on the right index and the right date
The formula is replayed with the published value of the index named in the contract, on the agreed revision date: not a quarter early, not on a neighbouring index.
Volume discount and year-end rebate checked
The cumulative base is recalculated as the contract defines it, then compared with what was actually deducted or credited.
Out-of-scope service flagged
A service billed but absent from the contractual scope is a discrepancy, even when its price looks consistent.
Every discrepancy then opens onto its evidence: the contract article and the invoice line, highlighted side by side, with the calculation replayed. Extracting the contract, meanwhile, is settled on page volume. Checking the negotiated terms is sized with you, 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 amendments, 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 an eighty-page contract extracted in full?
Yes, appendices included. The price schedule is rarely in the body of the contract: it sits in an appendix, sometimes in an attached spreadsheet, often altered by an amendment. All three are tied to the same agreement.
How are amendments handled?
An amendment is extracted like the contract, then dated: it replaces, extends or supplements specific line items. It is the effective date of each version that decides the price applicable to a given service.
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.
Is my data used to train models?
No, never. It is not shared between clients and stays hosted in France.
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
- Purchase order OCRThe commitment that ought to carry the schedule price, line item by line item.
- Supplier invoice OCRThe document the schedule finally makes checkable line by line.
- Long documents, down to the last pageTables running across thirty pages without losing a line.
- Automatic splitting of batched filesContract, appendices and amendments in one PDF, cleanly separated.
- Quote OCRComparing what is genuinely comparable: each line item returned separately, in the unit it was priced in.
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