MatterVerity — Reads a whole case file and traces every finding to its page.
Litigation intelligence for UK legal teams. It reads an entire case file and turns it into structured, source-verifiable intelligence (chronology, issues, contradictions, authorities and procedural obligations), with every finding traced to the page or message it rests on.
The problem
A contested matter can run to thousands of pages across statements, reports and message exports. The value is in the connections across the whole record, and every conclusion has to be checkable against its source.
What we built
Whole-matter analysis that links evidence, chronology and issues across every document, evidence-level provenance on each finding (document, page, paragraph or message range), confidence scoring, and human review built into the workflow.
What it does today
Chronology, issues, contradictions, authorities and procedural obligations from the whole record, each one a click from the page, paragraph or message it rests on.
— Under the bonnet
What makes a product like this hard to build
- 01
Whole-record context, not one document at a time
A case file can run to thousands of pages, far more than any model reads in one go. The system has to chunk, index and retrieve across every statement, report and message export, then reason over the connections between them without losing which document said what.
- 02
Provenance on every finding
Each conclusion has to point to a page, paragraph or message range a lawyer can open. That means carrying source positions through OCR, chunking, retrieval and generation, and rejecting any finding the model can't anchor to the record.
- 03
Messy legal inputs
Scanned bundles, handwritten annotations, WhatsApp exports and inconsistent date formats all need cleaning and normalising before they can be compared. Chronologies break if two documents describe the same day differently.
- 04
Contradictions and confidence
Spotting that a statement conflicts with a report is a comparison across documents, not a summary. Every finding carries a confidence score and a human review step, because a wrong 'contradiction' in a court bundle is worse than none.
- 05
Confidentiality
Client files are privileged. Access control per matter, audit logging and keeping documents out of any model training are built in from the start, not bolted on.
— Estimated build cost
What a build like this costs
£340,000–£500,000
Legal document intelligence platform · indicative range for a comparable build
Sized from the codebase we built: 361k lines of production code, 194k lines of tests, 218 database tables, 690 API routes.
A product like MatterVerity is a bespoke build, not a template: most of the cost is in the hard parts above, plus testing against real data before anyone relies on it. Where your version lands in the range depends on the factors listed here. You get a fixed price once scope is agreed — usually after a short discovery sprint. Hosting and AI model usage are billed at cost on your own accounts.
Get an estimate for yoursWhat moves the price
- Document volume and the OCR quality of the bundles
- Depth of source tracing (page level vs paragraph or message level)
- Which outputs are needed: chronology, issues, contradictions, authorities, deadlines
- Security and audit requirements for privileged material
- Integrations with case-management or document systems
Built with
- React
- TypeScript
- Express
- PostgreSQL
- Drizzle ORM
- Anthropic Claude
- OpenAI
Designed, built and run by our Manchester team. Want something similar? See our AI development services, AI automation and pricing.
