Building ChatGPT for Finance — Joseph Kim, OpenAI's Head of Financial Services

Key moments

The parts the AI thinks are worth your time, chosen from the creator's chapters. Click to jump.

1. 2:19Training financial reasoning into the modelShows how OpenAI thinks about teaching domain reasoning into a general model.
2. 5:29Why bundle financial data?Useful product logic on when to integrate data versus rely on connectors.
3. 6:39What can startups build?Directly points to startup opportunities on top of the same primitives.
4. 14:56Judgment, taste and the limits of accuracyExplains where model outputs still need human judgment and why that matters.

Chapters

From the creator's chapter list. Click to jump.

1. Cold open
2. 0:25Introducing Joseph Kim
3. 0:56ChatGPT for Financial Services
4. 2:19Training financial reasoning into the model
5. 3:31Codex adoption inside OpenAI
6. 4:27Building a finance product
7. 5:29Why bundle financial data?
8. 6:39What can startups build?
9. 9:18Safety and human review
10. 12:04The banking review chain
11. 12:36Financial mistakes and AI adoption
12. 14:56Judgment, taste and the limits of accuracy
13. 15:46Training models on human taste
14. 18:40Code quality as a taste benchmark
15. 19:04Investment ideas and moving up the analyst stack
16. 22:14Why financial evaluations are hard
17. 25:25OpenAI’s finance team is hiring

About this video

What does it take to build AI for finance on top of a general-purpose model? Joseph Kim, who leads work on ChatGPT for Financial Services at OpenAI, joins swyx and Richard at AI Engineer New York to discuss the decisions behind the product: training financial reasoning into a shared model, connecting market data, and building workflows financial professionals can review and trust. Joseph explains why OpenAI bundled financial data even when connectors already existed, what startups can build with the same plugin primitives, and why human taste and expert judgment are hard to measure. The conv