Health Records

Paper and scanned records need to become structured, searchable data – and frontline workers need to look them up quickly. This template shows an assistant that digitises records and answers questions over them.

How it works

1. Channels – a records clerk digitises; a health worker looks records up from an app.
2. Gateway & access – handles access and authentication.
3. AI agent (+ skills) – triggers OCR extraction, invokes the model, and retrieves and reads records on request.
4. OCR / extraction – reads scanned documents and writes structured output into the records service.
5. NAISH Hub – serves the OCR and language models on GPU. Sovereign AI core.
6. Data & dependencies – a scanned-docs store, a structured record store, a vector store for retrieval, and identity to gate access.

Where the Hub adds value

• Throughput – Hub GPUs process large digitisation backlogs far faster than manual entry.
• Lookup & grounding – retrieval lets workers query records in natural language, grounded in the real record store.
• Sovereignty – records and models stay on national infrastructure; see Governance & Risk.

Reusable building blocks