AI Can't Make Up a Number It Can't Look Up - Vinoo Ganesh, Kepler

Vinoo Ganesh, CEO and co-founder of Kepler and former head of business engineering at Citadel, explains why Wall Street still doesn't trust AI and how to build systems it can. He breaks wrong numbers into three causes (the wrong document, the wrong number from the right document, and models doing math) and shows Kepler's approach: code extracts every fact ahead of time and gives it an ID, the model plans and points to those IDs instead of writing numbers itself, and anything it can't resolve gets corrected or refused. He also covers wide-then-narrow search, scoping document access so firms never see each other's data, and the questions to ask any AI vendor about verification. Chapters: 0:00 Intro 1:38 Why Wall Street still doesn't trust AI 2:16 Hallucination vs alpha 3:14 Trust means someone to blame 4:17 Where wrong numbers come from 5:50 How Kepler is built differently 7:34 Pairing smart people with deterministic tools 9:07 Lookups can't get it wrong 9:26 Searching wide, then narrowing 10:46 Security for document search 12:25 How verification works 14:00 Verified, corrected or refused 15:40 Results from production 16:31 Questions to ask any AI vendor 17:29 We're out of the search phase Vinoo Ganesh Join us in SF for the second annual AIE CODE summit: https://aie.codes our last flagship summit of 2026. Applications live now!