ClauseGrade, a new independent testing firm built by credit practitioners, tests AI tools against a fixed set of synthetic lending tasks whose answer keys are computed in code. The suite has 40 tasks covering maintenance covenants, EBITDA add-back caps, borrowing bases, NAV facility tests, amendment chains and reporting deadlines. Nine are borrowing base cases, including aged and cross-aged receivables, concentration limits, a dilution trigger, an advance-rate step-down, an inventory sublimit and an overadvance. Fifteen of the 40 tasks contain no problem at all, so a tool that flags everything is marked down.
Each response is scored on four points: whether the tool spotted the problem, whether it got the number right, whether it escalated when a person needed to know and whether it cited the governing clause. Every task runs three times to measure consistency. A second checker, written separately from the main calculation engine, recomputes every answer key and the two agree on all 40 tasks.
Scorecards are private to whoever commissions them, and nothing is published without written consent. The fee is fixed per scope and does not depend on the result. ClauseGrade works with AI vendors that want outside evidence for their accuracy claims, and with lenders that want to test a tool before relying on it.
