OpenAI launches ChatGPT for Financial Services with data integrations
OpenAI released a financial services product built with Morgan Stanley and Evercore, integrating data sources like Dilupa and S&P. The company stated that third-party connectors had high error rates, necessitating in-house data integration.
OpenAI released ChatGPT for Financial Services, an enterprise product designed to combine consumer ChatGPT tools with firm-specific templates and integrated data sources like Dilupa and S&P. The product was developed with design partners Morgan Stanley and Evercore. Joseph Kim, head of ChatGPT Finance, stated that 95% of the product was built using standard plugin architecture and primitives available to external developers, while the remaining 5% involved specific model post-training. The team integrated data in-house because third-party MCP connectors exhibited error rates exceeding 40%.
The product assumes a human-in-the-loop workflow for financial analysts, leveraging existing hierarchical review systems in banking to mitigate risk. OpenAI is currently hiring product managers and subject matter experts to scale the team, and plans to make the product available to Pro plan users. Kim argued that bundling high-quality data directly into the product is necessary due to the high error rates of third-party connectors. "I really do think that there's going to be a billion dollar business built on top of a ChatGPT plugin," Kim said, suggesting the plugin architecture allows startups to build significant businesses on the platform.