Lingua LedgerDataNova 2026 · DN-AI-01

How it works

Extraction models are usually confident, but sometimes wrong. Lingua Ledger never relies on a single reading. It collects independent evidence for every field and turns it into a clear confidence score you can trust.

Document
PDF or image, any language
Vision reader
Reads the page image directly
OCR + text reader
Reads OCR text with a separate model
Arbiter
Settles reader disagreements using a crop
Validators
Checks sums, dates, IBAN, and VAT
Confidence
Five signals combined into one score
Accepted
Saves automatically
Review
A person checks it

Five signals per field

01

Reader agreement

The vision reader and the OCR + text reader work on their own. When they both return the same value, it is almost always right.

02

OCR confidence

We find the value on the page and use the OCR score for those words. This is also how we draw the boxes on the page.

03

Validators

Rule checks: line items add up to the subtotal, tax math is correct, dates make sense and are in order, and IBAN and VAT numbers are valid.

04

Self-reported confidence

Each reader reports how sure it feels. This is the weakest signal on its own, so it only gives the score a small push.

05

Arbiter verdict

When the two readers disagree, a vision model looks at a crop of the page to help score confidence. We still keep the merged text because replacing it barely helped accuracy in tests.

Memory and anomalies

Every posted document is saved to ledger memory by vendor. Future documents from that vendor are checked against it for duplicate invoice numbers, changed bank accounts, unexpected currencies, invalid tax rates, or future dates. Your reviews retrain the calibrator, so review work drops as the ledger learns.

Calibration

All signals are combined into a raw score and calibrated using isotonic regression tested on unseen documents. After that, a 0.9 confidence means the field is right about 90% of the time. The Evidence page shows the reliability curve before and after.

Evidence

Routing

Any field below the review threshold goes to a person, along with its box on the page and the evidence behind it. Raise the threshold for higher accuracy, or lower it to reduce human review. The Evidence page shows this trade-off on real data.

Workbench