Document Intelligence
Ingest, classify and index legal documents
Total Documents
—
Storage Used
—
Pending Review
—
Active Matters
—
Recently ingested
File types
By matter
Recent ingest jobs
Ingest Documents
Upload files or point to a local folder
Assign to matter (optional)
Drop files here, or choose an option below
PDF, Word, Excel, PowerPoint, Text, Images
Server-side folder scan
for network drives / mapped paths
Use this if your files live on a network share or mapped drive that the server can reach directly
Enable folder upload in Chrome
Chrome blocks folder selection on
http://localhost. Run generate_cert.py once to create a local SSL certificate, then restart the server. Chrome will allow folder uploads on https://localhost:8400.
venv\Scripts\python.exe generate_cert.py
Google Drive — direct download URLs
paste share links
Paste Google Drive direct download URLs, one per line. Use File → Share → Copy link, then convert to direct download format.
Document Index
All ingested documents
| Document | Type | Size | Matter | Status | Ingested | |
|---|---|---|---|---|---|---|
| Loading… | ||||||
Matters
Case groupings for document organisation
Loading matters…
Ingest Jobs
Folder scan history and status
PII Processing
Entity detection, obfuscation vault and review
Total documents
—
Processed
—
Pending
—
Entities found
—
Entity types detected
Recent processing log
No processing history yet
Entity vault
| Token | Type | Real value | Tier | Matched by | Detected |
|---|---|---|---|---|---|
| Run PII processing to populate vault | |||||
RAG Pipeline
Chunking → Embedding → Ready to query
Docs to chunk
—
Chunks created
—
Pending embed
—
Vectors in ChromaDB
—
Step 1 — Semantic Chunking
Splits obfuscated pages into parent (800 tok) + child (200 tok) chunks
Step 2 — Embedding
Sends child chunks to embedding provider → stores vectors in ChromaDB
Ask a Question
Agentic RAG — retrieval · reasoning · verification · citations
Top-K:
Ctrl+Enter to submit ·
Exhaustive: casts a wider net, finds more docs, slower
Brief Builder
Collect responses into structured legal briefs
Loading…
Select a brief from the list or create a new one
Document Search
Full-text keyword search · finds every document containing a term
Document Search vs Ask a Question:
Ask a Question uses AI to synthesise a legal answer — it finds the most semantically relevant chunks and cites them in context.
Document Search does a fast keyword scan — it finds every document containing the search terms, like Windows Search or Ctrl+F across all files. No AI involved. Use this when you need a complete list of all documents mentioning a specific district, docket number, or legal term.
Ask a Question uses AI to synthesise a legal answer — it finds the most semantically relevant chunks and cites them in context.
Document Search does a fast keyword scan — it finds every document containing the search terms, like Windows Search or Ctrl+F across all files. No AI involved. Use this when you need a complete list of all documents mentioning a specific district, docket number, or legal term.
Enter a search term to find documents
System Log
API and ingestion activity
Loading…