Structure the documents, keep the clinical decision
Referral letters, discharge summaries, lab reports and consent forms arrive as PDFs and scans. Munito AI gives healthcare teams open models and document AI on European infrastructure, with zero data retention by default.
Munito AI is infrastructure, not a medical device. Your application decides what reaches a clinician, and the clinician makes the decision.
From a document to a structured record
| Workload | Build with Munito AI |
|---|---|
| Document extraction | Convert letters, reports and forms into the fields your record system expects |
| Internal knowledge search | Answer a question of your staff from your own protocols, with the source passage |
| Administrative text | Draft and classify the correspondence around an episode of care |
| Coding support | Propose codes from a document and show the text behind each one |
Use Document AI for the pages, structured outputs for a fixed response shape, and the retrieval recipe for knowledge search. Choose from EU-hosted open models for each step of the workflow.
Patient data stays where your rules put it
Health data is special-category data under Article 9 of the GDPR, and each country adds its own hosting rules. France requires HDS certification to host personal health data, and Munito AI does not hold that certification. So run the platform inside the environment that already carries your approvals: your own cloud account, or on-premises.
Work that carries no patient data can run on European infrastructure that Munito AI operates, with the region pinned. Your team keeps one API across both, so a pilot on public documents moves into the clinical environment with no rewrite.
Munito AI does not train on customer data, and zero data retention is the default.
Give your data protection officer a short answer
Bring a DPA, a named deployment, a subprocessor list and a documented retention position to the review. Our SOC 2 audit and ISO 27001 certification work are in progress.
The security review guide collects those documents, and the subprocessor list names every company in the chain. Our engineers answer the technical questions of your data protection officer and your security team directly.
Prove it on documents you already hold
Start with one document type and a set of records your team already coded or summarized. Measure extraction accuracy, the quality of the source references, and the minutes each document saves. Review the output with the clinicians and the coders who will use it.
Move to dedicated capacity when the volume needs predictable throughput.
Talk to an engineer about your documents, your hosting requirements and your evaluation.