ChatGPT for Financial Services: What OpenAI’s Wall Street AI Can Do

OpenAI’s ChatGPT for Financial Services is a tailored version of ChatGPT Work for financial institutions. Announced on September 10, it combines GPT-6 Astra with built-in financial data, enterprise administration and tools intended for investment-banking and equity-research teams. Morgan Stanley and Evercore helped shape the product as design partners. 15

What it is designed to automate

The product targets time-consuming, document-heavy workflows often handled by junior banking and research staff. OpenAI positions it for developing research, financial models and customized client materials—not as an autonomous replacement for professional judgment. 56

Its intended uses include:

• Company and market research: synthesizing information across financial sources, including figures in filings, transcripts and other underlying material.
• Financial modeling: building, updating and analyzing complex Excel models while retaining existing formulas and workbook structures; OpenAI specifically highlights valuations, scenario analysis and tracing model drivers. 2
• M&A and acquisition analysis: supporting target or buyer screening, peer selection, financial analysis and the assembly of a presentation from those inputs.
• Pitchbooks and client materials: generating research notes, charts, documents and PowerPoint materials using firm-approved Excel, Word and PowerPoint templates and style guidance. 5

The practical promise is less manual gathering, reformatting and first-draft production. A bank remains responsible for the model, the advice, the client relationship and the decision to rely on any output.

Financial data and source-level citations

A major part of the proposition is putting premium financial information and AI reasoning in the same workspace. OpenAI says the product includes financial data and that it is designed to help teams create auditable work; reporting on the launch identifies Daloopa, PitchBook and LSEG as among the data providers. 12

The product also emphasizes granular citations: users should be able to trace a generated figure or assertion back to supporting source material rather than treating the response as an unexplained answer. That matters in finance, where an apparently small difference in a period, adjustment or definition can materially change a valuation or recommendation. 5

Citation visibility improves reviewability, but it is not a guarantee of correctness. Analysts should still check that a cited passage actually supports the claim, confirm units and reporting periods, and reconcile model outputs against the source data.

Integrations, connectors and controls

OpenAI says firms can centrally manage access and data connections through its enterprise platform. The broader financial-services offering highlights work in Excel and the ability to use existing financial-data relationships rather than forcing every institution into a separate research stack. 25

The launch materials describe integrations and connections with financial-data providers including S&P Global, FactSet and LSEG, alongside a wider connector ecosystem. Reporting also names Datasite, Box, Preqin and Intapp among connector examples. 78

That distinction is important: the available material supports Intapp as part of the connector ecosystem, not a separately documented, dedicated Intapp plug-in with a defined feature set. Firms evaluating the product should confirm their required data entitlements, information barriers, retention settings and connector behavior directly with OpenAI.

For regulated workflows, the product builds on ChatGPT enterprise controls such as identity and access management, role-based controls, configurable retention, encryption, audit-oriented logs and separate workspaces for information barriers. OpenAI also says customer business data is not used to train its models by default. 5

Who can use it—and what it costs

ChatGPT for Financial Services is available to eligible financial institutions through OpenAI sales or an existing account team. OpenAI had not publicly listed a standard seat price, minimum commitment or standalone SKU at launch. 39

That makes this an enterprise procurement decision rather than a consumer ChatGPT subscription. Institutions will need to evaluate the service alongside their market-data licenses, internal-data policies and review requirements.

How it compares with Anthropic’s financial-services push

Anthropic is the most visible direct competitor in this category. Its Claude for Financial Services offering includes agents for tasks such as pitch-material creation, earnings review, model work and compliance-oriented workflows.

Both vendors are pursuing the same broad opportunity: turn a general-purpose AI assistant into a governed workspace that can operate across proprietary data, office files and financial workflows. OpenAI’s positioning centers on GPT-6 Astra, integrated data, citations and a tailored ChatGPT Work experience; buyers should compare those capabilities with their current data stack, controls and document-production process rather than assuming either product is interchangeable. 5

The real debate: productivity tool or weakened apprenticeship?

The likely near-term effect is not the disappearance of bankers. It is the compression of repetitive work: collecting inputs, drafting research, updating workbooks, formatting slides and assembling first-pass client materials. That could give analysts more time for judgment, client preparation and deeper analysis.

The counterargument is that junior work is also training. Reconciling numbers, rebuilding a model, spotting a bad assumption and defending a conclusion are how analysts develop financial intuition. If teams accept AI-produced work without independently testing it, speed could come at the expense of that apprenticeship.

A sensible operating model is supervised use: require analysts to inspect citations, understand each model’s formulas and assumptions, reconcile key outputs and be able to explain the final recommendation. Source-traceable output can make that discipline easier, but the governance process—not the tool alone—determines whether automation strengthens or weakens financial reasoning.