OpenAI launched ChatGPT for Financial Services on Thursday, its latest move to build industry-specific tools on top of its core AI products. The announcement came as the company prepares for an initial public offering expected in 2027.
The product gives financial analysts access to datasets from Daloopa, PitchBook, and LSEG News directly inside ChatGPT — without needing separate contracts or subscriptions. OpenAI says bundling the data under one roof gets analysts information faster and in better formats.
The citation feature is the detail worth watching. Analysts can now pull 'granular citations' for any figure or claim, letting them trace data back to its original source as their analysis develops. For a Wall Street culture built on accountability and audit trails, that is not a minor upgrade.
OpenAI also said it is working to add data from S&P Capital IQ, Factiva, and Moody's to the product. Its latest model, GPT-6 Astra, is described as capable of retrieving information, running financial analyses, and synthesizing findings in sequence.
The company's stated ambition is direct: it plans to train its models to 'find, interpret, and use this data like we know the best analysts can.'
The finance industry has been moving toward AI adoption for some time. Morgan Stanley was an early adopter of ChatGPT, and leaders across Wall Street have been investing time and money into new AI use cases. OpenAI is now positioning itself to capture that demand at the platform level — owning the data layer, not just the model.
CFO Friar previewed the financial services announcement at a Goldman Sachs conference on Tuesday, naming banking alongside chip design, life sciences, personal finance, and health as priority verticals.
The bigger picture for founders: OpenAI is executing a classic enterprise playbook — land with a general-purpose tool, then build vertical depth that makes switching painful. Every data partnership it signs raises the cost for a bank to move to a competitor. The granular citations feature is not just a compliance convenience; it is a stickiness mechanism.
For founders building in fintech or enterprise SaaS, the signal is clear: the window to own a data layer before a foundation model absorbs it is narrowing. The companies that survive this cycle will be the ones that either go deeper on proprietary data than OpenAI can replicate, or ride the platform and build on top of it. There is no comfortable middle ground.
