OpenAI Introduces ChatGPT for Financial Services with Built-In Data and Granular Citations

ChatGPT for Financial Services: A Deep Dive

Okay, so OpenAI's launching a tailored ChatGPT service for financial services firms, "ChatGPT for Financial Services," as IT Home reported on September 11th. It was announced on September 10th (local time). The service looks pretty robust, built on a combination of financial data and GPT-6 Astra reasoning capabilities. I'm seeing this targets investment banking, equity research, and broader financial services teams. The core offering will help with research, financial modeling, and the creation of bespoke client materials. It is only available to eligible financial institutions.

The development process involved collaborations with Morgan Stanley and Evercore. The key pain points they're addressing are reliable data access and transforming analysis into usable, high-quality deliverables.

It's using built-in financial data from Daloopa, PitchBook, LSEG News, and Crunchbase. This data covers everything from earnings calls and financial statements to company fundamentals and private market data. OpenAI is handling the indexing and hosting, with traceable citations for everything, which is crucial.

They're also planning partnerships with S&P Global, LSEG, MSCI, Dow Jones Factiva, and Moody's to allow seamless integration of existing subscriptions. The tech also supports a wide range of connectors, with over 50 MCP connectors now supported, including Datasite, Box, Preqin, and Intapp.

The model itself, GPT-6 Astra, is designed to handle the nuances of financial documents. It can process the figures, tables, and footnotes. It will generate analysis, documents, spreadsheets, presentations, and even interactive charts, all based on the data.

They are including Excel, Word, and PowerPoint templates, which administrators will be able to manage via a dedicated admin page. This looks like it should facilitate creating things like valuation models, research reports, and pitch decks and roadshow materials that are consistent with institutional standards.

From a security standpoint, it's based on ChatGPT Enterprise. That includes SAML SSO, SCIM provisioning, and RBAC. They're making a point of saying that enterprise data won't be used to train the models by default, and that data encryption (both in transit and at rest) and data retention policies will be customizable.

For compliance, there's the OpenAI Compliance Platform for exporting workspace logs. They're also giving enterprises the ability to manage read/write access to skills and applications on a role-based basis, and support for information barriers by setting up multiple workspaces.

This seems to be accessible only to eligible institutions for now, but they're also providing API access for building custom applications.

IT Home, September 11 News: OpenAI announced the launch of ChatGPT for Financial Services on September 10 local time. This is a tailored ChatGPT service for financial services teams that combines built-in financial data with the reasoning capabilities of GPT-6 Astra, designed to help financial teams conduct research, perform financial modeling, and create customized client materials. The product is now available to eligible financial institutions.

According to the announcement, the product was co-developed by OpenAI with Morgan Stanley and Evercore, with an initial focus on investment banking and equity research use cases. OpenAI stated that during the collaboration, it identified key challenges faced by financial teams, including accessing reliable data and turning analytical results into high-quality work deliverables.

The service features built-in financial data from providers such as Daloopa, PitchBook, LSEG News, and Crunchbase, covering earnings calls, financial statements, company fundamentals, and private company information. This data is indexed and hosted by OpenAI, allowing users to trace the sources of specific figures and conclusions throughout their analysis.

OpenAI also plans to collaborate with organizations including S&P Global, LSEG, MSCI, Dow Jones Factiva, and Moody's to allow users to directly access licensed content through their existing financial data subscriptions. IT Home noted that OpenAI has also optimized MCP connectors for financial data providers, supporting more than 50 connectors including Datasite, Box, Preqin, and Intapp.

In terms of model capabilities, the service integrates GPT-6 Astra to process numbers, tables, and footnotes in financial documents, using them to conduct analysis, generate conclusions, and create documents, spreadsheets, and presentations. The product supports researching information across multiple data sources, tracking data changes over time, and generating interactive charts.

Financial institutions can also publish Excel, Word, and PowerPoint templates via a dedicated admin portal. Once administrators configure corporate templates and formatting guidelines, teams can transform analytical findings into valuation models, research reports, and pitch materials that comply with institutional standards.

Regarding security and governance, the product builds upon ChatGPT Enterprise's SAML single sign-on (SSO), SCIM user provisioning, and role-based access control. OpenAI stated that enterprise business data is not used to train models by default, data is encrypted both in transit and at rest, and administrators can define workspace data retention periods.

OpenAI also noted that compliance teams can export supported workspace logs via the OpenAI Compliance Platform for existing audit and investigation workflows. Enterprises can also manage read and write permissions for skills and applications based on roles, and enforce information barriers by establishing multiple workspaces.

ChatGPT for Financial Services is currently available only to eligible financial institutions. OpenAI stated that financial institutions and developers can also build applications tailored to specific financial workflows via its API.