Flagging policy

Risk-flag sensitivity: 70%

This case study models 500 vendor contracts, each scanned by an AI system for risky clauses (unlimited liability, unfavorable termination terms, missing indemnification). Higher sensitivity flags more clauses for attorney review but increases review workload; lower sensitivity risks a genuine risk clause slipping through unreviewed.

Portfolio outcome

Contracts flagged for review—

True risk clauses caught—

Attorney review hours—

Contracts by risk score (purple = flagged, red ring = true risk clause present)