AI Contract Analyzer · Applied AI · Document analysis
Making lease risk visible to non-legal stakeholders
An offline AI system that extracts, summarizes, and flags critical clauses in real-estate leases with 85% accuracy, cutting review time by 70% and giving non-legal stakeholders a clear view of risk.
70%reduction in review time
85%accuracy extracting and flagging critical clauses
- Role
- Designed and built the system.
- Tools
- Mistral 7B, Ollama, PyMuPDF, Python
How a lease becomes a risk summary
- 01Lease PDFUnstructured document
- 02ExtractPyMuPDF
- 03Summarize & flagMistral 7B via Ollama
- 04Risk viewCritical clauses for non-legal readers
The problem
Lease documents are long and unstructured. The clauses that carry real risk are easy to miss, and stakeholders without legal training had no quick way to see them.
Approach
- Extracted text from lease PDFs with PyMuPDF.
- Ran Mistral 7B locally through Ollama to summarize documents and flag critical clauses.
Outcome
Critical clauses were flagged with 85% accuracy, and review time fell by 70%.