OpenAI is making a more direct push into Wall Street workflows with ChatGPT for Financial Services, an industry-specific version of its enterprise offering designed for investment banks and financial institutions. The product was developed with Morgan Stanley and Evercore as design partners and uses GPT-6 Astra, cited by the company as its newest and most advanced model.
The stated goal targets a range of tasks that, in corporate finance, traditionally consume a significant share of analysts' and associates' time: gathering company information, reading quarterly earnings and earnings call transcripts, benchmarking comparable companies, crunching numbers, and turning analysis into presentations for clients and internal teams. OpenAI is therefore offering an assistant that does not stop at generating text, but is designed to link research, financial data, and outputs into formats ready for use in M&A transactions, valuations, and pitchbooks.
The move is notable primarily for where it fits into the process. Banks have already experimented with generative tools for narrow tasks, from document summarisation to internal search. ChatGPT for Financial Services, on the other hand, aims to integrate into the daily work that precedes a recommendation, a pitch, or a commercial proposal: the stage where quality depends on source selection, data consistency, the ability to explain market movements, and adherence to graphic templates and internal conventions.
From data to pitchbooks, without leaving the workflow
In the demonstration showcased by OpenAI, the platform was used to analyse a potential acquisition target. The system retrieved financial data from industry-standard sources and generated a PowerPoint presentation formatted according to a style guide pre-set by the bank. The example is significant because it illustrates a chain of tasks, rather than a single prompt: identifying a peer group, compiling pricing and information into a spreadsheet, verifying that the chart matches the data, and providing an explanation of the stock's performance.
In this line of work, automation is not measured solely by the speed at which slides are generated. A client presentation requires consistent data, verifiable sources, meaningful comparisons, and a structure that complies with institutional standards. This is also why OpenAI is focusing on teaching ChatGPT to conduct research and substantiate its conclusions using an approach close to that of an analyst.
The difference compared to ChatGPT Work, the enterprise product from which the new offering stems, lies primarily in native access to content essential in finance. ChatGPT for Financial Services can leverage data from LSEG, Daloopa, and Pitchbook, including information such as financial statements and earnings call transcripts. OpenAI has also enabled automated access to data subscriptions already held by users. For financial organizations, where proprietary or licensed information is an essential part of the work, this step is arguably more significant than the availability of a language model alone.
The product includes citations that allow users to trace back to source documents. This is an indispensable feature in an industry where a figure entered into a table, a valuation multiple, or a claim regarding a company’s performance must be verifiable. The risk of inaccurate or unsupported answers remains one of the structural limitations of generative systems, and it is magnified when the output ends up in materials intended for financial decisions or external stakeholders.
This is not an announcement of the end of analysts
Framing the initiative as an outright replacement for junior bankers would, however, be a simplistic interpretation. Analysts and associates do not merely produce spreadsheets and presentations: they work under the guidance of senior teams, understand the context of client relationships, verify assumptions, coordinate revisions, and manage processes where judgment, confidentiality, and professional accountability cannot be delegated to software.
The most tangible change concerns the makeup of the workday. If the system succeeds in reducing the time required for initial research, data gathering, and standard formatting, teams will be able to focus more on verification, interpretation, and deal management. At the same time, banks may rethink how junior tasks are structured—tasks that for decades served as both an operational necessity and a training ground for those climbing the investment banking ranks.
This shift will demand rigorous controls. Market data and corporate disclosures are subject to licensing terms, usage restrictions, and continuous updates. Banks will need to determine which sources are permitted, how audit trails are preserved, who approves outputs, and when the system cannot be used. While inline citations and connections to existing subscriptions address part of the challenge, they do not eliminate the need for internal procedures and human review.
Then there is the question of analytical quality. A model can surface data and build comparisons quickly, but selecting peers, defining assumptions, and interpreting market movements require context. The value of these tools will therefore depend less on generating visually polished pitch decks and more on their reliability in making reasoning transparent, flagging uncertainties, and enabling precise source verification.
The race to dominate finance software
The announcement comes amid broader competition for the enterprise market, where OpenAI faces rivals such as Anthropic and Google. Anthropic had already introduced Claude for Financial Services, a dedicated offering for the industry. Vertical specialization is one of the primary avenues through which model providers are seeking to turn experimental interest in generative AI into production deployments: convincing a regulated institution takes more than a general-purpose chatbot; it requires integrations, trusted data, guardrails, and tools compatible with existing workflows.
For OpenAI, finance is also a major commercial proving ground. Sarah Friar, the company's chief financial officer, told investors in August that the enterprise business generated more revenue than the consumer segment, which grew following ChatGPT's debut in 2022. The new product reinforces this direction, while the market eyes the prospects of a potentially major stock market listing for the company.
Nick Turley, vice president of product at OpenAI, indicated that Financial Services will not be an isolated case: the company plans to release tailored solutions for other sectors as well. It is a sign of the next phase of generative AI in the enterprise. Following general-purpose assistants, the competition is shifting toward the ability to integrate into professions with specific languages, databases, workflows, and obligations.
For investment banks, the arrival of ChatGPT for Financial Services does not eliminate the issue of oversight, nor does it make human expertise superfluous. However, it can compress the time between an initial request and a first draft of documented analysis. If the data integration and promised controls hold up in real-world use, the pressure will fall primarily on the manual, repetitive tasks that have defined a substantial part of entry-level work on Wall Street until now.
