Ras Mic (“Professor Ras Mic”)

Engineer at Tempo (an AI startup) and a frequent technical-explainer guest on Greg Isenberg’s podcast. Known for clean, non-technical breakdowns of AI infrastructure topics — particularly the LLM → tools → MCP evolution.

Channels

  • Guest appearances: Greg Isenberg’s podcast and similar
  • X/Twitter: linked from podcast show notes
  • Day job: Tempo (AI startup, builds agentic tooling)

Content in This Wiki

  • MCP, clearly explained — Ras Mic’s full non-technical walkthrough of MCP: standards analogy, the three-stage evolution of LLM capability (raw → tools → MCP), the four-component architecture, and Anthropic’s “3D chess” play of pushing server implementation onto service providers.

Key Ideas

  • LLMs by themselves are “incapable of doing anything meaningful” — the value comes from connecting them to tools
  • Glue code between LLMs and tools is the dominant cost of AI startup engineering — MCP is the standardization that finally addresses it
  • Anthropic externalized the integration cost: by putting MCP servers in the hands of service providers, they made every MCP client more capable without paying for it themselves
  • “Standards are why engineers can communicate” — programming progress is gated on shared standards, not new ML

See Also