Skip to content

CASE-FILE MODE

Index the entire case file. Ask in natural language.

A 10-second query over 170+ documents, audio recordings and videos from your case. Every answer cites the exact source: document, page, timestamp.

What you can index

Native PDFContracts, judgments, deeds, expert opinions
Scanned PDF (OCR)Legacy archives digitised on the fly
AudioWitness statements, hearings, recorded meetings
VideoAny case-relevant recording

Designed for case files of 170+ documents with no performance degradation.

How it works

01

Index

Upload the case corpus once. The system chunks each artefact, vectorises it via high-dimensional semantic embeddings and stores it in a semantic knowledge base bound to the file.

02

Query

Ask anything in plain language — about timelines, contradictions, clauses, witness statements. No keyword grammar required.

03

Answer

The system replies with mandatory literal citations: [Source 1] Lease.pdf · p. 7 or [Source 3] Hearing 12-Mar.mp3 · 14:32–14:58. The model can never assert anything outside the corpus.

Example

  • "When did the principal witness testify and what exactly did they say?"
  • "Are there contradictions between the plaintiff’s expert and the defence’s?"
  • "Which clauses of the contract govern early termination?"

[Source 1] lease.pdf · p. 7 · [Source 3] hearing-12-mar.mp3 · 14:32–14:58

Operational capabilities

🔌 Live ingestion connectors

Bring material into the matter without manual uploads

Three connectors are live: a Box integration pulls files straight from your Box account, an Outlook add-in files messages and attachments from your inbox, and an email-to-matter alias auto-ingests inbound emails and their attachments into the matter. Material lands in the case file ready to index.

⏱ Real time savings

300 pages → 10 seconds

Locating a specific datum in a 300-page case file goes from hours of manual search to a 10-second query.

🔗 Auditable trail

Every citation is expandable

Click any source to see the literal fragment of the original document, alongside the semantic relevance score.

🛡 Corpus-grounded

Out-of-corpus → not answered

The model works exclusively over the documents you indexed. If the answer isn’t there, the system says so explicitly instead of inventing.

🔒 GDPR by design

PII Gatekeeper before the model

Names, IDs and contact data are tokenised before any fragment reaches the language model. Confidentiality is enforced at the infrastructure layer.

Where it’s used

  • Trial preparation with extensive case files
  • Documentary due diligence in M&A
  • Coherence analysis across multiple related contracts

Ready to index your first case?

Available now in Nexus Legal. Open the platform and use the 📁 Case-File Mode button next to the document upload area.

Open Case-File ModeRequest a guided demo