Skip to content
Deep-OCR.
DE

For documents that deserve discretion

Confidential documents.
Structured data.

OCR and structured extraction for your software: contracts, receipts and case files become readable, searchable and analyzable. Processed only in the EU, with clear deletion periods and no model training.

For law firms, tax advisors, medical practices and the teams building their software.

A considered approach to data
SAMPLE DOCUMENT

Invoice

EX-2026-0142
Example office · Berlin

Consulting1.000,00 €
VAT 19%190,00 €
Total1.190,00 €
Structured outputJSON
invoice_number
EX-2026-0142
total_amount
1190.00
currency
EUR
Confidentiality, built into the process.Processing only in the EUNo model trainingDefined deletion periods

01 / Confidential by design

Your documents. Your professional secrecy. Our diligence.

Sensitive information needs more than accurate text recognition. What matters is where it is processed, how it is used and when it is deleted.

Processing only in the EU

The platform runs in Germany; text recognition and extraction use AI models operated in data centers within the EU. Your documents are not transferred to third countries, including the USA.

Defined deletion periods

Files are deleted after processing, results after a short retrieval window. Deep-OCR is not a document archive.

No model training

Your documents and extraction results are not used to train or improve AI models.

Contract documents for professional secrecy

Data processing agreement under Art. 28 GDPR, technical and organizational measures, and a confidentiality undertaking under Section 203 of the German Criminal Code. Our subprocessors are bound by the same obligations.

02 / Applications

Where discretion is part of the job.

For professional practices and any organization working with confidential documents. Integrated into the software you already use.

Law & advisory

More time for the case.

Extract parties, dates and terms from contracts. Make scanned files available as searchable full text for your own systems.

Contracts · Case files · Correspondence

Tax & accounting

From receipt to workflow.

Turn invoices, receipts and bank statements into structured fields for your accounting and document workflows.

Invoices · Receipts · Bank statements

Healthcare

From scan to patient record.

Make referral letters, findings, discharge reports and referrals usable: as searchable full text for the patient record, or as structured fields such as diagnosis, date and treating facility for your practice or hospital software.

Referral letters · Findings · Discharge reports · Billing

03 / The data lifecycle

A clear process. Right up to deletion.

Temporary processing with a defined retrieval window. Here is what happens between upload and deletion.

  1. Upload

    Send your document via the API. The upload is temporarily stored encrypted.

  2. Extract

    Extract structured fields or full text. The upload is deleted after processing.

  3. Retrieve

    Retrieve the result within one hour and pass it to your own workflow.

  4. Delete

    The result is deleted after the retrieval period. Deep-OCR is not a document archive.

Account and billing data have separate retention periods. Read the privacy policy

Anonymization and redaction

Only pass on what is needed.

Dear Ms Musterfrau,

regarding your matter AZ 2026/0142 of 12 March 2026 we enclose the requested documents. Please direct any questions to m.musterfrau@example.org.

name
removed
case_number
pseudonymized
date
kept

Recognize personal data such as names, addresses, dates of birth and case numbers, and remove or redact it before the result is passed on. Downstream systems, analyses and AI assistants then receive only what they actually need.

Which fields are removed, pseudonymized or kept is defined per workflow, so professional secrecy is protected without losing the information you work with.

04 / Integration

Fits into your workflow.

Send a document, retrieve its processing status and use the result in your own application. Structured JSON for defined fields, or Markdown for full text.

One document. Two steps.REST API
01

Submit a document

POST /v1/extractionsfile + document_type → 202 + id
02

Poll until processing is complete

GET /v1/extractions/{id}queued → processing → completed / failed
Failed jobs are reported explicitly.

05 / Pricing

Start with your first document.

Evaluate the extraction with representative sample documents. Choose the plan that fits your volume.

All plans include full API access. All prices exclude VAT.

Special requirements or larger volumes?

Good to know

Clear answers. Before you start.

Is AI involved, and where does it run?

Yes. Text recognition and extraction use AI models. The platform itself runs in Germany, the models in data centers within the EU. Your documents are not transferred to third countries, including the USA, and are not used to train models. The processors involved are listed in our data processing agreement and bound by the same confidentiality obligations.

Is Deep-OCR a ready-to-use office application?

Deep-OCR is an API for integration into existing software and workflows. Your development team or software partner connects it. The portal includes a playground for evaluating extraction on sample documents.

What does deletion mean in practice?

Uploads are temporarily stored encrypted and deleted after processing. Completed or failed results can be retrieved for one hour, then are deleted. Jobs that have not been processed within two hours expire. Account and billing data have separate retention periods; see our privacy policy.

Does Deep-OCR automatically anonymize documents?

Yes. Personal data such as names, addresses, dates of birth and case numbers can be recognized and removed, pseudonymized or redacted before the result is passed on. Which fields this applies to is defined per workflow. Note that removing names alone does not make a document anonymous, so we review the relevant fields with you.

What about professional secrecy?

We provide a data processing agreement under Art. 28 GDPR, a list of technical and organizational measures, and a confidentiality undertaking under Section 203 of the German Criminal Code. The same confidentiality obligation is in place with our subprocessors, so the chain of confidentiality stays intact. Request the documents and we will send them for your review.

Which documents and outputs are supported?

Deep-OCR processes PDFs, images and common office documents such as DOCX. For supported document types such as invoices, contracts or bank statements we return structured JSON; for everything else, the full text as Markdown. Formats, limits and billing are described in the API documentation.

Let's talk about your documents

Confidential deserves a closer look.

Together, we can review your document types, integration and data-protection requirements.