AI Contract Decomposition Engine
An AI-powered tool that ingests unstructured contract documents and decomposes them into structured data — extracting obligations, parties, dates, risk clauses, and key commercial terms automatically.
Solo Builder
2024
The Problem
Legal and procurement teams spend hours manually reading contracts to extract structured information. This is expensive, slow, and error-prone — particularly at scale across large contract portfolios.
The Approach
Built an LLM-powered extraction pipeline with a structured output schema covering parties, obligations, termination clauses, payment terms, SLAs, risk indicators, and renewal dates. Designed the UX around a confidence-scored output model — every extracted field shows an accuracy indicator with the source text for human verification.
Architecture / System Design
Document ingestion via PDF parsing → chunking strategy optimised for legal text structure → multi-pass extraction using chain-of-thought prompting → structured JSON output validated against schema. Built review UI for human-in-the-loop correction. Deployed as a standalone web app.
Outcomes
Reduces contract review time from hours to minutes. Deployed live. Forms the basis of the Contract Risk and Gap Assessment platforms (also built).