Scanning gives you an image; our capture process gives you the data. Our operators run every file through Sentinel Capture, our own in-house intelligent document processing platform, which classifies, reads and validates each document — turning boxes of paper into accurate, structured data that flows straight into your systems. Sentinel Capture runs on two engines: our own in-house engine, trained only on blank forms and never on your data, which handles many jobs entirely within our UK systems; and a UK-hosted enterprise engine, under strict no-training terms, for the rest. Because the platform is our own, we can offer what off-the-shelf tools cannot: pseudonymisation before processing, a taxonomy tailored to each client, and a full audit trail behind every field — all checked by our own people in our ISO 27001 UK facilities.
A scanned document is just a picture of paper until someone keys the information off it. Our intelligent document processing reads the content itself — supplier names, totals, dates, references — and turns it into structured data, validated and ready to use.
Our capture process does the heavy lifting; our people keep it accurate. Anything the system is unsure of is flagged for a quick human check, so you get speed without sacrificing the audit trail.
We've moved beyond the template-based capture platforms this industry has leaned on for a decade. In side-by-side testing on real client documents, our current capture engine materially outperformed the legacy tools it replaces — reading messy, handwritten and unstructured documents accurately with no templates and no per-layout setup.
Read supplier, totals, VAT and line items from any layout and post them into your finance system.
Extract responses from questionnaires and structured forms into clean, analysis-ready datasets.
Classify and index personnel files, contracts and right-to-work documents, sensitive data handled securely.
Inbound post classified and routed to the right person, team or workflow automatically.
Surface key terms, dates and parties so renewals and obligations never slip through the cracks.
Index and extract data from medical and case records to NHS-aligned standards.
Documents are scanned or received digitally, and our Sentinel Capture process identifies what each one is.
The engine reads the fields that matter from each document, whatever the layout.
Confidence scoring flags anything uncertain for a quick human review.
Validated data flows into your finance, HR, EDRM or records platform.
A typical back-file arrives as hundreds of undifferentiated pages — typed, handwritten, stapled, out of order. Here's what our engine does with it, with identifiers pseudonymised before a single page is read.
The engine reads each page the way an experienced records clerk would — recognising what it is, who it concerns and when it happened — then rebuilds the file digitally: classified by document type, ordered into a patient chronology, and indexed with the fields your teams actually search by.
Anything ambiguous — an unreadable date, a page that could belong to two documents, a possible misfile — is flagged for a DBS-checked reviewer rather than guessed at.
Mixed bundles split automatically into GP notes, discharge summaries, referrals, test results and imaging reports — no separator sheets, no manual prep.
Every entry dated — including handwritten notes — and ordered into a single timeline, so a clinician reads a history, not a pile.
NHS number, demographics, medications and doses, allergies, diagnoses and clinician names captured as structured, searchable fields.
Repeated copies and superseded versions identified and flagged — the record you keep is the record that matters.
Output filed into your EDRM or records structure with metadata on every document — retrievable by any field, not just a name.
Confidence scoring routes unreadable pages, missing dates and possible misfiles to trained reviewers — never silently guessed.
The same engine handles legal case files, HR records, invoices and public-sector archives — the document types change; the discipline doesn't.
The same accredited team that stores and scans your records now reads them for you.
Sentinel Capture reads any layout — tables, handwriting, poor copies — with no templates or per-format setup.
Confidence-scored extraction with human review on anything uncertain — no blind automation.
All processing runs inside our accredited UK environment — your data stays in an accredited chain.
Outputs into finance, HR, EDRM and records systems, or our secure cloud portal.
We've been trusted with the UK's most sensitive records since 1977. Our capture process runs under that same discipline — engineered around UK GDPR from the first page in to the last field out.
Where required, personal identifiers are stripped or pseudonymised inside our ISO 27001 UK facilities before any content reaches our capture engine — and re-linked to your records only once the data is back inside our secure environment.
Your documents are processed under strict data-processing agreements: they are never used to train AI models, and retention is contractually controlled — including zero-retention processing where your governance requires it.
Encrypted in transit and at rest, processed in isolated batches, with a unique audit reference for every job — the same chain-of-custody discipline we apply to physical records.
A data-protection impact assessment for every engagement, documented lawful basis, records of processing and international-transfer safeguards where applicable — plus support for subject-access requests over processed data.
Confidence thresholds route anything uncertain to DBS-checked reviewers working in our UK facilities. No blind automation — speed from the machine, judgement from people.
Every extracted field links back to its source document and page, so an auditor can walk from a number in your system to the exact page it came from.
Our capture process reads your documents. It never decides their fate — retention, release and destruction always sit with people, under your policies. Read our full Data Protection Statement.
They flex around our schedule — even working weekends to hit our data deadlines. For any data-capture challenge, they'd be our first recommendation.
Send us a sample set and we'll show you exactly what our capture process can classify and extract — then scope it for your volumes.
Book a capture demoIt goes beyond scanning. Rather than just producing an image of a page, our Sentinel Capture process classifies each document, reads the content itself — supplier names, totals, dates, references — validates it, and delivers structured data straight into your business systems.
Scanning creates a digital image of a document; data capture reads the information off it. We do both — digitising your paper and turning the content into accurate, structured data your systems can use.
Sentinel Capture, our in-house platform, runs on two engines: our own in-house engine — trained only on blank forms and sample documents, never on your data — which handles many jobs entirely within our UK systems; and a UK-hosted enterprise engine, under strict no-training terms, for the rest. Anything the system is unsure of is flagged for a quick check by our own people.
No. Our in-house engine is trained only on blank forms and mock-ups, never on real client data, and the enterprise engine runs under strict data-processing agreements that prohibit training on your data — with zero-retention processing where your governance requires it.
Yes, by design. Documents are prepared in our ISO 27001-accredited UK facilities, personal identifiers can be pseudonymised before processing, and every engagement is backed by a data-protection impact assessment, a documented lawful basis and a full audit trail.
Whatever your process needs — supplier, totals and VAT from invoices; responses from forms and surveys; key terms, dates and parties from contracts; and structured fields from HR, clinical and case records. Any information trapped on paper or in unstructured documents.
Our capture process does the heavy lifting and anything it is unsure of is flagged for a quick human check — human-in-the-loop validation gives you speed without sacrificing accuracy or the audit trail. Every extracted field links back to its source page.
Yes — we read supplier, totals, VAT and line items from any invoice layout, with no templates, and post them into your finance system after validation.
A mixed patient file can be automatically sorted into document types — GP notes, discharge summaries, referrals, test results, imaging — ordered into a dated patient chronology, and read into structured fields such as NHS number, demographics, medications, allergies, diagnoses and clinician names. Anything uncertain is routed to a trained human reviewer, and identifiers are pseudonymised before processing under our GDPR-first architecture.
Yes — we scan and classify your incoming post and route it to the right person, team or workflow automatically. See our Digital Mailroom service for how it works.