Hacker News
Kepler Finance: How 147 Financial Firms' Distrust of AI Led to a Verifiable Research Platform Built on Claude
Anthropic published a case study on Kepler, a financial research platform built by ex-Palantir engineers who interviewed 147 financial firms and found universal distrust of AI output. Kepler's architecture uses deterministic infrastructure as a trust layer with Claude as the reasoning engine, enabling analysts to ask plain-English questions and receive instantly verifiable answers traced to source documents. At 41 points and 22 HN comments, it signals enterprise AI adoption gated on auditability, not capability.
↳ Follow the thread