Pond Street Ledger
RobinStreet and more, read daily
Every story this desk has filed under Claude, most recent first.
A decision model takes content and a question and returns a typed answer with a calibrated confidence, not prose, in a single forward pass. TypeSafe AI announced one on 15 September with a $40 million seed round led by DCVC. Levanto Labs published its own, Sage, to npm on 13 July, and Tripcite has been running it on every self-serve audit since 27 July: Claude writes 30 candidate questions, Sage ranks them, and the audit measures the top 14. In the test that earned it the job, Sage put all 14 real queries above all five planted bad ones.