Anthropic took its biggest swing yet at financial services on Tuesday, using an invite-only event in New York to roll out a battery of pre-built finance agents, deep Microsoft 365 connectors, a native Moody's app inside Claude, and a fresh set of data partnerships designed to make Claude Opus 4.7 the default reasoning layer for global banks.
The announcements cap a 48-hour blitz that started with a $1.5 billion joint venture with Blackstone, Hellman & Friedman and Goldman Sachs, and now extends into the actual workflows analysts, underwriters and credit officers run every day.
Roughly ten purpose-built finance agents
According to Fortune's reporting on the event, Anthropic unveiled around ten purpose-built agents covering pitchbooks, earnings analysis, credit memos, underwriting, KYC, month-end close, statement audits and insurance claims. Production users named in the announcement include JPMorganChase, Goldman Sachs, Citi, AIG and Visa.
Claude Opus 4.7 is positioned as the brain behind these agents. Anthropic says the model now leads Vals AI's Finance Agent benchmark with a 64.4% score and tops the GDPval-AA evaluation — claims the company is leaning on heavily as it pitches risk-averse Wall Street buyers.
Moody's, Verisk and a wall of new data connectors
The data side of the announcement is arguably more strategic than the model itself. Moody's has embedded its full platform into Claude as a native app for credit analysis, while new connectors from Verisk, Third Bridge, Fiscal AI, Dun & Bradstreet, Experian, GLG, Guidepoint and IBISWorld plug Claude directly into the reference data and expert networks bankers already pay for.
For a Wall Street buyer, that combination matters: agents become useful only when they can pull authoritative data without analysts babysitting copy-paste flows.
Microsoft 365 add-ins go GA
Anthropic also moved its Microsoft 365 integration to general availability, letting Claude operate across Excel, PowerPoint, Word and Outlook in a single workflow. The add-ins are designed to let Claude read a model in Excel, restructure a deck in PowerPoint and draft the cover note in Outlook without leaving Office — the same surface area Microsoft Copilot has been defending for two years.
A direct shot at OpenAI's enterprise narrative
CFO Krishna Rao framed the push bluntly: "Enterprise demand for Claude is significantly outpacing any single delivery model." That is the subtext of the entire May 5 announcement. Anthropic isn't selling banks a chatbot — it is bidding to become the system of record for high-stakes financial reasoning, with a joint venture for distribution, agents for workflow, Moody's for data and Microsoft 365 for the surface.
The risk is execution. Pre-built agents in regulated industries live or die on accuracy, auditability and the ability to integrate with legacy mainframes that Wall Street has, in many cases, refused to retire. But after months of being shut out of the Pentagon's classified networks, Anthropic now has a very clear answer to where its enterprise growth is supposed to come from — and it's wearing a suit.
