Anthropic debuts 10 AI agents for finance workflows

Anthropic on Tuesday released 10 prebuilt AI agent templates aimed at financial firms to automate tasks such as building pitchbooks, screening know-your-customer files and running month-end close processes.

The templates are available as plugins for Claude Cowork and Claude Code and as cookbooks for Claude Managed Agents. Each template combines skills that embed domain knowledge, connectors that provide governed access to relevant data, and subagents-smaller Claude models that handle subtasks such as selecting comparable companies or checking methodologies.

The agents are organized into two groups. Research and client coverage templates include a pitch builder that creates target lists and drafts pitchbooks, a meeting preparer that assembles client briefs, and an earnings reviewer that reads transcripts and flags changes relevant to investment theses. Finance and operations templates cover general ledger reconciliation, month-end close checklists, audit support and KYC compliance screening.

Anthropic extended Claude’s integration with Microsoft Office so the model can work across Excel, PowerPoint, Word and Outlook with context carrying between applications. The company added data connectors from Dun & Bradstreet, Fiscal AI, Guidepoint, IBISWorld, SS&C IntraLinks, Third Bridge and Verisk, and launched a Moody’s MCP app that provides credit ratings and data on more than 600 million public and private companies.

Anthropic paired the templates with Claude Opus 4.7 and reported a top score of 64.37% on Vals AI’s Finance Agent benchmark. Users must review and approve Claude’s outputs before they are shared with clients or filed.

At the announcement event, Anthropic CEO Dario Amodei warned that advanced AI tools can uncover software vulnerabilities faster than organizations can fix them, describing a ‘moment of danger’ and suggesting there may be a six- to 12-month window to address tens of thousands of flaws. Anthropic limits access to its Mythos model; partners have used it to find and patch hundreds of vulnerabilities.

The release comes amid industry moves to use AI to increase efficiency. Companies including Meta, Block and Coinbase have tied recent layoffs in part to plans to automate work with AI.

Anthropic described the templates as intended to speed routine work for analysts and compliance teams while keeping human oversight in the final review process.