OpenAI aims ChatGPT for Financial Services at junior banker work
Built with Morgan Stanley and Evercore as design partners, it researches companies and builds pitchbooks — the apprenticeship at the bottom of every bank.

OpenAI has built a version of ChatGPT aimed squarely at the work Wall Street gives its youngest employees: researching companies, analysing financial data and producing the pitchbooks that investment banks run on.
ChatGPT for Financial Services, unveiled on Thursday, is a tailored edition of the company's enterprise product ChatGPT Work, developed with Morgan Stanley and Evercore as design partners. It runs on GPT-6 Astra, OpenAI's most advanced model.
The job it is designed to do
The target is not a vague notion of productivity. It is the specific, decades-old apprenticeship at the bottom of every investment bank — the recent graduates known as analysts and associates, employed to research deals and build presentations.
"We're effectively teaching ChatGPT to research like an analyst and back up its conclusions like an analyst as well," said Nick Turley, OpenAI's vice president of product, at a briefing announcing the product.
The second half of that sentence is the more interesting claim. Producing a plausible financial summary is something general-purpose models have managed for a while. Producing one whose conclusions can be traced back to sources a managing director will accept is the part that has kept this work with humans.
Why banks, and why now
The choice of design partners is what OpenAI is leaning on for credibility: Morgan Stanley and Evercore are the two firms named as having shaped the product before launch.
The commercial logic behind the launch is equally clear. OpenAI has spent much of the last year competing for business customers against Anthropic and Google, and the enterprise market is where the money now is: finance chief Sarah Friar told investors in August that the company's enterprise business accounted for more revenue than its consumer business — the side that took off after ChatGPT launched in 2022.
This also arrives as OpenAI prepares for what is widely expected to be a blockbuster initial public offering, and a demonstrable enterprise franchise is a more legible asset to investors than consumer subscriptions.
It is not first to the sector. Anthropic announced Claude for Financial Services, its own tailored offering for Wall Street, last year.
What it means for the bottom rung
The work OpenAI describes — researching deals, building pitchbooks — is what the industry has used analysts and associates for over decades, by its own account of the role. Automating those tasks does not obviously remove the need for people who have done them. It does remove the thing they were doing while they learned.
That is the structural question this product raises and which no vendor announcement can answer: if the research and the pitchbook are generated, what does a first-year analyst do instead, and where does the next generation of managing directors acquire its judgement?
A tool that produces the analysis is a different proposition from one that speeds up producing it.
The claim that has to hold
The constraint on a tool like this is accountability rather than capability: an analysis that is right most of the time is of limited use if nobody can say which part is wrong.
Turley's framing — research like an analyst, and back up its conclusions like an analyst — is a direct answer to that objection. Whether it holds in practice, on live deals, under compliance review, is what the design partnerships with Morgan Stanley and Evercore exist to establish.
OpenAI has not published error rates, nor described how the product cites and evidences its conclusions in enough detail to assess them independently.
What has not been said
Pricing has not been announced. Nor has the company said how widely the product is being made available, whether it is restricted to institutional customers, or what data-handling arrangements apply to the confidential material an investment bank would necessarily feed it — the single most consequential question for any bank considering adoption.
Turley told reporters that OpenAI plans further releases, which suggests financial services is a template rather than a one-off. Which sector gets the next tailored edition has not been said.



