AI OCR with bounding boxes and grid extraction
OCR and IDP with geometry: fields, locations and archive PDF from one API.
View feature →Features
Document AI features — one API, one account
PaperOffice AI bundles OCR with bounding boxes and sandwich PDF, anonymization, e-signatures (SES), WORM/GoBD, knowledge graph, PDF API, workflows/HITL and voice in one credit model — EU hosting, MCP-ready.
At a glance
- 300+API/MCP tools
- 35UI languages
- 69ready document types
- Full-text +semantic search
Made for real business. Not for hype.
Được tin tưởng bởi các công ty hàng đầu trên toàn thế giới
Feature pages
Eight core capabilities — each with API context and EU hosting.
Automatically anonymize documents (PII detection + redaction)
Detect and redact personal data — directly in the archive with an audit trail.
View feature →E-signatures via API: signing links, OTP, mobile
SES signatures with evidence retention on the document — PaperOffice does not currently offer QES.
View feature →Tamper-evident archiving: WORM, legal hold, audit trail
Audit-proof filing with WORM/GoBD focus, legal hold and EU hosting.
View feature →Knowledge graph across all documents
Link documents via content and relationships — searchable for teams and agents.
View feature →PDF API: compress, convert, PDF/A, protect
Create and process PDFs in the same account as extraction and archive.
View feature →Workflows, HITL and approvals via API
Approvals and review on the document — without a media break after extraction.
View feature →TTS, voice clone and document podcasts
Over 840 voices, 39 direction markers, pronunciation lexicon — and every audio becomes a searchable document.
View feature →FAQ
What is a Document AI feature via API?
A Document AI feature is a single capability (e.g. OCR, anonymization, e-signature) reached through the same API and credit model — without a separate product account.
How do Features differ from Extract?
Features describe platform capabilities (OCR, WORM, voice). Extract lists document-type pipelines (invoice, ID, delivery note) with fields and endpoints.
Is PaperOffice MCP-ready?
Yes. The same capabilities are reachable via REST and the MCP server — one account, one token, one balance.
Where is data processed?
In PaperOffice EU infrastructure on owned hardware — without mandatory US hyperscalers.
What does usage cost?
Billing is per page or credits; Elite typically starts around €9 per 1,000 pages depending on the pipeline. See the pricing page for details.
Is an SDK required?
No. The API is REST-first; optionally MCP is available for Claude, ChatGPT, Cursor and other clients.
What does human-in-the-loop mean?
Ambiguous cases can be reviewed and corrected on the document. Corrections stay on the object — including an audit trail.
Can features be used individually?
Yes. Many capabilities can be used standalone; the MCP-native AI-DMS optionally bundles them in one platform.
How does a quick start work?
Create an API key, pick an endpoint from the docs, or use MCP Connect — typically integrated in minutes.
Is there a feature matrix versus competitors?
Yes. /compare/ provides a verified matrix; unverified cells stay “n/a”.
