Mistral LLM x Law Hackathon.

Pivot helps lawyers compare a case with past court decisions and identify missing facts that could change the legal analysis.

Discover the project

The facts that could change a case.

Built as a team at the second LLM x Law Hackathon in Paris, hosted by Mistral AI and Stanford Law School, Pivot helps lawyers explore conflicting precedents. Starting with a question and supporting documents, it compares the case with earlier decisions and surfaces the missing facts that could shift the analysis.

Role: Frontend development & interaction design
Type: Hackathon
Stack: Next.js / React / TypeScript / Python / FastAPI / Mistral AI / MCP
Status: Hackathon prototype
Year: 2026

Pivot’s case prompt with a question about the employment status of couriers at the fictional UrbanShift platform

Pivot’s loading screen showing progress while preparing a case for analysis

From the legal question to case preparation.

Explore the case, one fact at a time.

Extract
Mistral reads the documents and extracts facts with supporting quotes, leaving missing information marked as unknown.
Compare
A separate Python calculator weighs the precedents, identifies material differences, and estimates where the case leans.
Simulate
Change a fact, such as real-time location tracking, from Unknown to Yes and compare the updated estimate with the original case.
Inspect
Open source excerpts behind the facts and precedents, and see which decisions were retained or excluded from the analysis.

Making the analysis easier to explore.

I built the Next.js and TypeScript frontend, from document upload and preparation states to the analysis dashboard. My work focused on the editable fact matrix, contextual evidence panels, and the comparison between the original case and a simulated change.