Anthropic just made its most aggressive move into enterprise software yet. On May 5, 2026, the company held an invite-only "Briefing: Financial Services" event in New York and announced 10 purpose-built AI agents designed specifically for banks, insurers, asset managers, and fintech firms. The launch came with expanded data partnerships, full Microsoft 365 integration, and a joint venture backed by Blackstone, Hellman & Friedman, and Goldman Sachs. Anthropic AI agents for financial services are now available today, and the scale of this rollout signals that the race for Wall Street is no longer theoretical.

What the 10 Anthropic AI Agents for Financial Services Actually Do

Anthropic organized its 10 new agents into two categories that mirror how financial firms actually operate: Research and Client Coverage, and Finance and Operations. Each agent is a reference architecture built on three components: skills (domain-specific instructions and knowledge), connectors (governed access to real-time financial data), and subagents (specialized Claude models that handle specific subtasks like comparables selection or methodology checks). Firms can customize every agent to their own policies, modeling conventions, and approval workflows before deployment.

The Research and Client Coverage agents include a Pitch Builder that assembles comps models in Excel and drafts pitchbooks in PowerPoint, a Meeting Preparer, an Earnings Reviewer, a Model Builder, and a Market Researcher that tracks sector and issuer developments and flags items for credit and risk review. On the Finance and Operations side, the agents cover Valuation Review, General Ledger Reconciliation, Month-End Close, Statement Audit, and KYC Screening, which assembles entity files, reviews source documents, and packages escalation files for compliance teams. According to Anthropic's announcement, tasks that once took analysts hours or days can now be completed in minutes with greater consistency and fewer errors.

The Data Connector Ecosystem

AI agents are only as useful as the data they can access, and Anthropic has built out one of the most comprehensive financial data connector ecosystems of any AI platform. Existing connectors already include LSEG, S&P Capital IQ, Morningstar, PitchBook, FactSet, MSCI, Chronograph, and Daloopa. New additions announced at the event include Dun & Bradstreet, Experian, Verisk, Third Bridge, GLG, Guidepoint, Fiscal AI, and IBISWorld, all integrated under governed access controls with per-tool permissions and a full audit log.

The headline partnership is with Moody's, which is embedding its full platform into Claude as a native app. That gives analysts the ability to pull credit ratings and risk data on more than 600 million companies without leaving the Claude interface. Markets reacted immediately: FactSet shares fell as much as 8.1% on the day of the announcement, Morningstar dropped more than 3%, and both S&P Global and Moody's saw significant selling pressure. Investors are clearly reading the data connector strategy as a direct threat to the business models of established financial data providers.

Microsoft 365 Integration

Anthropic also announced that Claude is now embedded across Microsoft 365, covering the four tools financial professionals spend most of their working day inside. In Excel, Claude builds financial models directly from filings and data feeds, audits formulas across linked workbooks, and runs sensitivity analyses. In PowerPoint, it drafts pitchbooks and presentations that update automatically when the underlying numbers change, and in Word it edits credit memos against a firm's own templates.

Claude for Outlook is arriving soon and is designed to act as a chief of staff, triaging inboxes, scheduling meetings, and drafting responses in the analyst's voice. The key capability across all four platforms is persistent context: an analyst who begins a financial model in Excel does not have to re-explain it when the work moves to PowerPoint or Word. According to Anthropic's product page, analysts can also assign Claude work tasks by text or voice through Dispatch in Claude Cowork, with finished work ready for review while they are away from their desk.

The Model Powering It All

All 10 agents run on Claude Opus 4.7, which Anthropic positions as its most capable model for financial work. According to TechRadar, Claude Opus 4.7 scored 64.37% on Vals AI's Finance Agent benchmark, ahead of GPT-5.5 at 59.96% and Gemini 3.1 Pro at 59.72%. The model also leads the GDPval-AA evaluation for economically valuable knowledge work. These benchmarks matter most to the enterprise buyers making procurement decisions, and Anthropic is clearly building its product narrative around them.

It is worth noting what that 64.37% benchmark also means in plain terms: the model gets things wrong more than a third of the time on the most rigorous finance-specific evaluation available. Anthropic addresses this directly by designing every agent with mandatory human review before any output goes to a client, gets filed, or is acted on. The value is not autonomous execution but eliminating the hours of prep work that precede the expert judgment calls that still require a human.

Who Is Already Deploying These Agents

The financial services adoption story is not hypothetical. According to Reuters, roughly 40% of Anthropic's top 50 customers are now financial institutions, and financial services represents Anthropic's second-largest industry by enterprise revenue, behind only technology clients. The client list includes Goldman Sachs, Visa, Citi, AIG, and JPMorganChase, all of which have Claude running in production workflows.

FIS, which sits inside the core banking infrastructure of roughly 12% of the world's banks, is co-developing a Financial Crimes AI Agent with Anthropic that compresses AML investigations from hours to minutes. BMO and Amalgamated Bank are among the first deployers, with broader availability planned for the second half of 2026. Carlyle has adopted Claude across investment analysis, operations, and portfolio management, citing its agentic reasoning and coding capabilities as core to delivering value across the firm.

The Business Strategy Behind the Launch

This launch does not exist in isolation. Fortune reported that Anthropic CEO Dario Amodei disclosed at the event that the company projected 10x revenue growth in Q1 2026 and instead saw 80x actual growth. Anthropic is also reportedly exploring a funding round that could value the company at more than $900 billion, and an IPO could come as early as this year. The financial services push is the enterprise revenue narrative that supports both of those things.

The joint venture with Blackstone, Hellman & Friedman, and Goldman Sachs is designed to push Claude into the portfolios of mid-sized banks, manufacturers, and health systems at a scale that no software vendor has previously had access to. The two-track strategy, giving large institutions the tools to configure and run agents themselves while using the joint venture to embed Claude directly into mid-market operations, is an attempt to capture the full spectrum of enterprise financial services in a single coordinated move. That is not a product launch. That is an infrastructure play.

What This Means for Financial Firms

The question for financial services firms is no longer whether AI agents will become standard operating infrastructure. The early adopters across banking, private equity, and insurance are already running these tools in production, and the data connector ecosystem Anthropic has assembled makes it significantly easier to deploy without a major procurement cycle. Firms that already pay for FactSet, S&P Capital IQ, Moody's, or LSEG have a relatively low-friction path to connecting Claude to the data they are already using.

The compliance architecture is also designed to align with regulatory expectations. Governed access controls, per-tool permissions, full audit logs, and mandatory human approval before any action are all built into the agent framework. Every agent is designed to draft and surface work for review, not to act unilaterally. For compliance teams evaluating whether these tools fit inside their regulatory obligations, that architecture is the right starting point for the conversation.