Solutions · Digital Foundations
Make your records
answer questions.
Document Intelligence is the end-to-end version of our records work: assess what you hold, digitise what matters, extract and structure it, then make it retrievable and answerable under your permissions.
Assess.
Digitise.
Make answerable.
What it is
Not a scanner. Not a chatbot.
A scanner produces images. A chatbot produces sentences. Neither produces an archive you can govern. Document Intelligence is the pipeline between them, and the pipeline is the product.
The seven stages run in order because each one depends on the one before it. Classification applied to poor extraction produces a well-organised archive of the wrong thing. An assistant built on unclassified records answers from whatever it happens to find.
What you receive
Deliverables, not deliverables-shaped slides.
Each item below is something you keep, whether or not the engagement continues past that stage.
A records assessment
What exists, in what state, how much of it matters, and what should not be carried forward at all.
A digitised archive
Captured at a quality that survives extraction, with the pages and volumes logged so gaps are visible.
An extraction and quality report
What OCR handled cleanly, what it did not, and what a person needs to correct. Measured on your documents, not a benchmark.
A classification and metadata scheme
Built around how you need to retrieve records, not around how they happen to be filed today.
A permissioned repository
One place to find a record, with access defined by role and an audit trail of who opened what.
A knowledge assistant, where scoped
Answers drawn from approved records, with the supporting passage shown and permissions respected.
An ingestion workflow
Because records keep arriving, and a one-time digitisation project becomes an abandoned archive without a route for new material.
Where it fits
Organisations with records that matter and cannot be found.
- Professional practices holding years of client files, correspondence and signed agreements.
- Clinics and service businesses with patient or client records across paper and disconnected systems.
- Property, legal and institutional archives where a document must be found and defended years later.
- Organisations whose institutional knowledge currently lives in the memory of two or three people.
The limits, stated plainly
What this will not do.
- It will not make a scan a legally authoritative replacement for an original document. That is a decision about records policy and regulation, made with you, not a feature.
- It will not extract every document perfectly. Handwriting, stamps, poor scans and unusual layouts have error rates, and we measure them rather than quote a general figure.
- It will not decide who is allowed to see what. Permissions come from your organisation and your obligations, and we implement them.
- It will not maintain itself. Ongoing ingestion needs an owner inside the organisation, and that is agreed before launch.
We are building a Document Intelligence demonstration. It is not published yet, and it will not be presented as live before it passes the same testing the other demonstrations did.
Document Intelligence
Start with one archive,
not the whole organisation.
Pick the records that cost you the most to find. We will assess that collection and tell you what a full implementation would involve.