Pre-process
Every page typed, every doc routed
Per-page classification with confidence sends invoices, statements and IDs to the right schema — first time, cited.
Routing you can audit
01
Per-page labels
Not one guess per file — a label + confidence for every page.
02
200+ doc types
Invoices, statements, pay stubs, IDs, BOLs, ACORD forms and more.
03
Custom routes
Pin types to your schemas; unknowns queue for labeling.
Classification coverage
Types route automatically where confidence clears your threshold.
Run your first check- 01
Invoices & receipts 0.99
Headers, line items and totals vocabularies per locale.
- 02
Bank statements 0.96
Any bank, any format — multi-page tables kept together.
- 03
IDs & PoA 0.97
Passports, national IDs, utility bills and bank letters.
- 04
Freight docs 0.95
BOL, CMR, POD and rate confirmations across languages.
- 05
Unknowns review
Below threshold pages queue with suggestions — label once, remembered.
From mixed file to typed docs
01
Split finds pieces
Or upload single docs — classification runs either way.
02
Each page labeled
Type + confidence per page, majority wins per piece.
03
Routed to schema
Invoices to AP, statements to spread — unknowns wait for you.
Right schema, first time.
Classification before extraction — no more statements parsed as invoices.
See batch queueRelated use cases
High-90s on common types (invoices, statements, IDs) with per-page confidence. Low-confidence pieces queue with suggestions.
RO/TR/DE/BG plus Cyrillic scripts proven — the interpreter reads labels, not dictionary templates.
Yes — pin labels to your schemas and set per-type thresholds. Unknowns queue for one-click labeling.
No surcharge — extraction meters per page as usual. Batch adds the 50% discount on top.
Route every page right
Labels with confidence.
A mixed AP inbox sorts itself: invoices to AP, statements to spread, unknowns to review.