Observed arrival · 2026-10-22
ClauseGrade tests whether credit AI gets the clause right
An independent testing firm evaluates AI tools on private-credit work such as covenant checks, borrowing bases, and NAV facilities.
○Open
⊠Login
$Paid
†Ads
✦Pretty
●Pro
◎Niche
◉Human
⚑Risk
ƒJS
- For
- Private-credit teams evaluating AI tools
- Worth noticing
- Fifteen of the 39 tasks contain no problem, testing whether a tool can avoid flagging clean cases.
Field notes
The test materials are designed by hand, while expected answers come from a calculation engine using exact decimal arithmetic. A separately written checker reportedly recomputes each key with exact fractions, and the two must agree before the suite is used. The homepage says the fictional agreement language follows formulations common in credit agreements filed publicly with the SEC, rather than copying any single agreement.
Observed signals
Read the marks
Editorial observations of this landing page, not a rating.
○OpenPublic substance visible
$PaidCommerce or pricing visible
✦PrettyNotable craft visible
●ProPolished or operationally mature
◎NicheUnusually specific use
One card from the complete issue