Descriptions:
Peter Yang sits down with Ryan Wiggins, head of product at Mercury, for a live demonstration of Mercury’s MCP (Model Context Protocol) integration and a first look at Wiggins’s personal AI productivity system built on Claude Code. Mercury MCP, available in the Claude app store, gives users natural-language access to their banking data — querying spend patterns, identifying tax opportunities, and summarizing financial activity — using a read-only connection to Mercury’s API. Wiggins also reveals on-camera that Mercury is releasing a CLI tool in the coming weeks for developers who want lower-latency, lower-context-overhead access than the MCP channel provides.
The second half of the conversation shifts to Wiggins’s “second brain” setup: a Claude Code knowledge base assembled from nearly five million words of documents, meeting notes, and project artifacts accumulated over five years in his role. He uses a tool called Pod to capture meeting audio, which feeds back into the system and surfaces relevant context during live calls. The setup gives him what he describes as a queryable, persistent version of his own professional memory.
The discussion also covers the strategic dynamics of MCP as a distribution layer — Wiggins draws an analogy to the App Store, noting that both OpenAI and Anthropic have converged on MCP as the standard third-party integration surface, making it a meaningful go-to-market channel for companies like Mercury that want to reach users inside AI assistants.
📺 Source: Peter Yang · Published April 22, 2026
🏷️ Format: Interview
