Anthropic did something uncharacteristic.
The company that has always focused on "building models" quietly open-sourced a repository called financial-services. Not a model, not an API — but a complete set of Claude workflow templates for the financial industry.
In one week, stars went from under 10K to 20K. People in finance are clearly more excited than the tech crowd.
What's Inside
This isn't a demo or toy project. It contains:
- Research report generation: Claude analyzes earnings reports, news, and industry data to generate structured research reports
- Compliance review: Automatically checks if financial documents meet regulatory requirements
- Client communication templates: Email and report generation for financial clients
- Risk management framework: Claude-based risk assessment workflow
All templates are deployable Python code, not documentation descriptions.
Why Anthropic Is Doing This
My read: this isn't about building a product. It's about anchoring industry best practices.
When financial institutions say "we want to use AI," what they lack isn't model capability — the Claude API is right there. What they lack is concrete answers to "how exactly should AI be used in finance."
Anthropic gave them a reference implementation. This means:
- Financial institutions don't need to start from scratch
- The path to using Claude gets standardized
- Other model companies wanting to enter finance need to match this quality
It's an ecosystem play, not a product play.
Testing: Is It Usable?
I tried the research report generation template.
Good parts: Clean code structure, detailed comments, professionally written prompts (clearly written by someone who knows finance). Generated reports have proper formatting and accurate data citations.
Caveats:
- Default config requires Claude API key, and costs aren't trivial (financial docs are long, token consumption is high)
- Some templates assume access to Bloomberg or similar financial data sources — not free
- Compliance prompts are US-regulation focused; domestic use needs adjustment
Who Should Care
Should look:
- Fintech companies planning AI implementation
- Quant teams wanting AI-assisted research
- Anyone interested in "how do LLMs actually work in traditional industries"
Can skip:
- Individual investors (not useful for personal use)
- Developers who don't work in finance
GitHub: github.com/anthropics/financial-services | Python | 20K stars | Anthropic official