Two years ago I started a newsletter tracking freshly raised VC funds. The core insight was simple: founders were pitching investors who had already deployed their capital. They needed to know who had dry powder right now, not who raised a fund in 2021.
That newsletter became Fund Momentum. Today it tracks 970+ active funds with live GP signals — who's bullish on what, how to pitch each partner, deployment status.
The newsletter had 43K+ subscribers. Good. But I kept seeing the same friction: founders researching investors manually, copying data into spreadsheets, guessing at GP preferences. This was exactly the kind of repetitive research that AI agents should be doing.
So I built an MCP server. Now the same intelligence that founders read in the newsletter is queryable by Claude, LangChain, CrewAI, or any AI workflow at €0.01 per call.
Subscriptions don't work for agents. An agent running 3,000 queries at 2am doesn't care about monthly resets. So I built a credit system. Buy once, use forever. Every response tells the agent exactly how many credits remain.
When credits hit zero, the MCP returns a structured error with a buy_url. The agent surfaces this to a human who tops up. Clean handoff.
Developer inbound after the GitHub README. Founders using it in Claude Desktop for investor research before fundraising calls.
Getting developers who build fundraising tools to discover FM before they build their own scraper. That's the distribution challenge.
Try it: https://fundmomentum.vc/for-agents
GitHub: https://github.com/schneidavie/fundmomentum