Lease Contract Red-Flag Analysis
Ranking 179 lease agreements by severity of automatically extracted red flags using the ALeaseBERT dataset
About This Analysis
Many organizations need to sort large collections of legal documents by latent attributes—for example, which contracts pose the greatest risk to the tenant. DocETL pipelines can express this kind of semantic ranking in a few declarative steps.
This demo takes 179 lease agreements and automatically ranks them by the severity of "red flags"—provisions that are potentially dangerous for the lessee (early-termination clauses, uncapped fee escalations, etc.). The pipeline first extracts candidate red-flag sentences for each contract, then ranks the contracts by severity. Contracts at the top contain the most egregious red flags; those at the bottom are relatively benign.
Processing details: 179 lease contracts • gpt-4.1-mini • $4.24 total cost
Note: There is no "ground truth" ranking for severity. The AI's assessment is based on general legal principles and may not capture all nuances or jurisdiction-specific issues.
Explore the Output
Select any contract to read it with highlights. Click a red flag on the right to jump to its location in the text.