Building a case brief without re-reading every PDF

LegalMemo is built for litigation teams in Indian courts. Upload the case PDFs, and it gives back a structured brief with a timeline, conflicts, admissions, and a clean set of facts, each one tied to the source page it came from.

What the work normally looks like

Building a case brief is hours of work, and most of it is re-reading. An associate goes through every PDF in the matter, pulls out the facts, marks the dates, lines up the statements, and writes the brief the partner can actually use.

Every step is slow and easy to get wrong. Miss a date and the timeline is off. Miss a quote and a fact has no source. Miss a contradiction and the strategy is built on shaky ground.

What LegalMemo does

The team uploads the case PDFs. A five-stage background pipeline takes over.

• Stage 1. Document ingestion. Every page is read, including scanned and rotated ones. A dual-track OCR pipeline using pdfplumber, Tesseract, and Claude catches what single-tool OCR misses.
• Stage 2. Fact extraction. Each fact is pulled out with its date, the people involved, and the exact source quote it came from in the PDF.
• Stage 3. Timeline. The facts are arranged into an ISO-date chronology, so the team can see how the case unfolded.
• Stage 4. Conflict detection. The pipeline finds places where statements contradict each other across documents.
• Stage 5. Admissions detection. It also catches places where a party contradicts their own earlier position, which is often the most useful kind of contradiction in court.

The associate then reviews the output, resolves ambiguities, and exports a structured 15-section brief as a PDF.

Why every fact has a quote

A brief that cannot be sourced is a liability. Every fact LegalMemo extracts comes with the exact quote and a pointer to the page it came from. Click any fact and the source viewer opens the original PDF at the right page, with the bounding box drawn around the quote.

That is the part that makes it usable in actual practice. The partner gets a thorough brief, and every claim in it is one click away from the original text.

What the team gets back

Briefs that used to take a week take a day. The associate spends their time on the parts that need a lawyer's judgement, conflict resolution and strategy, not on careful reading. The partner gets a brief that is more thorough, with the receipts already attached.

The case file also becomes searchable. Months later, when someone needs a specific fact or a specific contradiction, it is already there, indexed and sourced.

The stack underneath

LegalMemo runs on FastAPI with a React front end. Celery and Redis handle the background pipeline. Supabase stores the PDFs and the structured outputs. The LLM layer is provider-agnostic. The whole thing is built to be deployed on the team's own infrastructure if the firm requires it.