OpenAI is taking a more vertical approach to enterprise AI with ChatGPT for Financial Services, a tailored ChatGPT Work experience built around GPT-6 Astra, financial data, and common banking deliverables. The company says the offering is designed for teams working on research, valuation models, earnings analysis, and client materials rather than for general-purpose office assistance.
Data is the product’s starting point
The offering includes datasets from Daloopa, PitchBook, LSEG News, and Crunchbase. OpenAI says the coverage includes earnings transcripts, financial statements, company fundamentals, private-company information, and related research inputs. The data is indexed and hosted on OpenAI infrastructure, which the company says lets it improve retrieval, latency, and how evidence is presented.
That packaging is meant to remove a familiar friction point in financial workflows: analysts often have to search across subscriptions, connect systems, and reconcile figures before they can start reasoning about a company. OpenAI says the product can show the tables and passages behind a figure or claim, giving bankers a way to inspect the evidence while an analysis is being developed.
Built around the work bankers already produce
OpenAI says the initial design partnership with Morgan Stanley and Evercore helped steer the product toward investment banking and equity research. The company describes a workflow that can navigate figures, tables, and notes; perform financial analysis; and synthesize the result into documents, spreadsheets, and slides.
Administrators can publish Excel, Word, and PowerPoint templates through a dedicated admin page. With a firm’s templates and style guides configured, teams can use the output for valuation models, research notes, and pitchbooks without rebuilding the same formatting by hand. The announcement presents this as a way to move from analysis to a reviewable work product inside the firm’s existing conventions.
Governance for sensitive workflows
The product builds on ChatGPT Enterprise identity and administration features. OpenAI lists SAML single sign-on, SCIM provisioning, role-based access controls, encryption in transit and at rest, configurable workspace retention, and supported compliance-log exports. It also says business data is not used to train its models by default and that firms can manage skills and apps by role.
Those controls matter because the most valuable financial workflows often contain confidential client information or material non-public information. They do not, by themselves, make an AI-generated analysis investment advice or a substitute for a bank’s review process. The bank still owns the obligation to validate data, assumptions, calculations, and client-facing language.
Availability and the competitive signal
ChatGPT for Financial Services is available to eligible financial institutions through OpenAI’s sales channel. The launch puts model capability, licensed data, output templates, and governance in one package, a strategy that could make adoption easier for large firms while narrowing the room for teams to mix and match providers.
Newsroom analysis
The announcement is less about a new model than about controlling the path from trusted inputs to auditable work. For finance teams, the practical test will be whether citations remain reliable across messy documents, whether templates preserve human review, and whether the system’s permissions and retention settings fit each firm’s information barriers. OpenAI’s product page describes the intended workflow; it does not independently verify performance or regulatory outcomes.