Axiom — semantic search for law firms
Semantic search platform that lets law-firm associates query 30 years of internal precedent, contracts, and matter files in natural language.
- OCR pipeline (PDF + scans)
- Vector embedding store
- Search UI + result ranking
- Matter-scoped RBAC
- Audit logging
What they needed to solve.
Axiom's customers (top-50 UK and US law firms) were running fee-earner research across PDF folders, archived emails, and Sharepoint. Average research time per matter was 6 hours. Internal precedent was unsearchable.
How we approached it.
A semantic search platform with OCR ingestion, vector embeddings of every paragraph, and citation-grade results with provenance back to the original document. Custom RBAC keeps every matter visible only to authorised teams.
Design decisions.
- AaSöhne — UI
- AaGT Sectra Text — citation body
- AaSöhne Mono — clause numbers
Citation-grade UI — every result shows a paragraph excerpt with its full document path and access trail. No 'AI summary' without source.
- Citation provenance for every result
- Snippet-first results, full doc on demand
- Audit trail of every search per fee-earner
What changed.
PDF folders + Sharepoint + email archives, no semantic search, no audit trail, 6 hours research per matter.
Semantic search across 30 years of firm history, full audit trail, citation-grade snippets, matter-scoped RBAC.