Anthropic launches AI agents for finance and insurance | Keryc
Anthropic introduced a package of AI agents designed to speed up the slow, repetitive tasks in banking, asset management, and insurance. Can you imagine finishing month-end close or preparing a pitchbook in days instead of weeks? The goal is to turn those tasks into supervised, repeatable workflows.
What Anthropic announced
The company published ten ready-to-use templates that automate key tasks: from building pitchbooks and financial models, to reviewing statements, reconciling ledgers, and doing KYC. Each template ships as a plugin for Claude Cowork and Claude Code, and also as a cookbook for Claude Managed Agents.
The idea is simple: teams that today take months to integrate AI can put Claude to work in days.
New agent templates (practical summary)
They present templates grouped by function. They're configurable so you can adapt them to your internal policies, models, and risk controls. Here’s a condensed list so you can quickly see what’s available:
Research and client coverage
Pitch builder: lists targets, calculates comparables and assembles pitchbooks.
Meeting preparer: prepares briefs ahead of calls.
Earnings reviewer: reads transcripts and filings, updates models and flags meaningful changes.
Model builder: creates and maintains models from filings and feeds.
Credit, risk and compliance
Market researcher: monitors sectors, news and research for risk alerts.
KYC screener: builds files, reviews documents and prepares escalations.
Finance and operations
Valuation reviewer: checks valuations against comparables and methodologies.
General ledger reconciler: reconciles accounts and calculates NAV.
Month-end closer: runs a close checklist, prepares journal entries and reports.
Statement auditor: reviews statements for audit readiness.
Each template combines: instructions and domain knowledge, connectors to governed data and specialized subagents.
How they're used in practice
You have two main modes:
Plugins inside Claude Cowork or Claude Code: the agent works alongside the analyst using the apps already on their desktop. For example, the Pitch agent can deliver an Excel model, a PowerPoint deck and a coverage email ready in Outlook.
Claude Managed Agent: runs autonomously on the platform for long workflows, like overnight processes or book closes that span several hours. It includes long-running sessions, tool-level permission controls, a credential vault and a full audit trail.
In both cases, a human reviews and approves before any output is sent to clients or archived.
Integration with Microsoft 365
Claude now works directly in Excel, PowerPoint and Word via add-ins, and Outlook is coming soon. What does that mean day-to-day?
In Excel: it creates and audits models, connects feeds and runs sensitivity analysis.
In PowerPoint: it generates decks that update when the numbers change.
In Word: it edits credit memos according to your bank’s templates.
In Outlook: it acts like a chief of staff that triages email, schedules meetings and drafts replies in your voice.
Context flows automatically between apps, so you don't need to re-explain a model when you turn it into a presentation.
Connectors and data ecosystem
Agents perform best with high-quality data. Anthropic connects Claude to platforms like S&P Capital IQ, PitchBook, Morningstar, LSEG and firms’ internal systems, all under governed controls.
New connectors have been added: Dun & Bradstreet, Fiscal AI, Financial Modeling Prep, Guidepoint, IBISWorld, SS&C IntraLinks, Third Bridge and Verisk. Moody's also released an MCP app that brings ratings and data on hundreds of millions of companies.
Connectors and MCP apps let Claude use real-time data and provider-specific tools inside the interface.
Real use cases and adoption
Several banks, asset managers and insurers are already using Claude for front, middle and back office tasks. There are examples where agents compress AML research from days to minutes, and where investment teams improve meeting prep by turning execution time into idea time.
Firms report broad adoption: from engineering teams using Claude Code to analysts integrating Claude in Excel to speed up due diligence and modeling.
Risks and control
Anthropic emphasizes governed data access and traceability. Agents log every tool call and every decision so compliance and tech teams can audit the process. Still, human oversight remains key: agents assist, they don't replace professional responsibility.
How to get started today
The templates are available in the financial services marketplace. They can be used as plugins on paid plans of Claude Cowork or Claude Code, or as Managed Agents on the public beta platform. The new connectors and Moody's MCP app are also available to joint customers on paid plans.
Anthropic also offers webinars and sales support to show practical implementation.
This move shows how AI stops being a promise and becomes an everyday tool in finance: not to replace human judgment, but to free up time and raise the quality of analysis. Ready to try an agent that handles the heavy lifting and leaves you the intellectual part?
