Our Take
These are packaged workflows, not breakthrough capabilities, but the Office integration and managed deployment options address real adoption barriers.
Why it matters
Finance teams can now deploy Claude on high-value tasks without custom engineering, and the Microsoft Office integration means work stays in familiar tools instead of requiring new interfaces.
Do this week
Finance teams: Test the pitch builder and KYC screener plugins in Claude Cowork this week to measure time savings on your actual deal flow.
Ten finance templates plus Office suite integration
Anthropic released ten pre-built agent templates targeting specific finance workflows: pitch building, KYC screening, earnings review, financial modeling, general ledger reconciliation, and month-end closing. Each template packages task instructions, data connectors, and subagents into deployable units that work as plugins in Claude Cowork and Claude Code, or as standalone managed agents.
Claude now works directly inside Microsoft Excel, PowerPoint, Word, and Outlook (coming soon) through add-ins. Context carries automatically between applications, so work that starts in Excel can move to PowerPoint without re-explanation. The pitch builder agent, for example, can produce a comparables model in Excel, draft a pitchbook in PowerPoint, and prepare a cover note in Outlook from a single target list input.
New data connectors include Dun & Bradstreet for business identity verification, SS&C IntraLinks for deal room access, Third Bridge for expert interviews, and Verisk for insurance data. Moody's launched an MCP app providing credit ratings and data on 600 million companies (company-reported).
Packaged deployment addresses adoption friction
The templates solve a practical problem: most finance AI projects stall in custom development. By shipping reference architectures with connectors and approval workflows included, teams can deploy on real work in days rather than months (per Anthropic). The Office integration keeps work inside familiar tools instead of forcing interface changes.
Customer adoption signals validate the approach. Walleye Capital reports 100% Claude Code adoption across their 400-person hedge fund (company-reported). Multiple asset managers and banks are already using Claude for model building, research synthesis, and operational workflows according to the customer quotes provided.
The managed agent option handles multi-hour processes like deal closes with persistent sessions, credential management, and full audit logs. This addresses compliance requirements that block many finance AI deployments.
Start with high-frequency, low-risk workflows
Begin with the pitch builder and KYC screener templates since these have clear time savings metrics and defined approval processes. The Excel integration works best for model updates and sensitivity analysis where you can verify outputs against existing work.
For managed deployments, focus on operational workflows like month-end close and GL reconciliation that run on schedules and have structured outputs. The audit logging satisfies compliance review requirements that manual processes often lack.
Avoid starting with client-facing work until you've validated output quality on internal tasks. The templates include approval workflows, but your firm's risk tolerance and review standards still apply to any client deliverables.