When AI Gets Your Types Wrong
A Pragmatic Solution
How I built a TypeScript definitions MCP server to help AI tools understand and mock complex types correctly, turning a testing frustration into a practical solution.
Read essay →Essays on technology, leadership, and the human systems that make or break great teams.
Recent Essays
The Rise of Generalist Leaders
How Agentic Tools Are Reshaping Engineering Leadership
How agentic tools are democratizing technical capability and reshaping engineering leadership, creating organizations led by generalists who synthesize across domains.
Read essay →What I'm thinking about:
- • How do we maintain system understanding when AI writes more of our code? Hard-won insights came with well-fought battles, not just reading the code. This feels like it matters now but will it mater later (in a few years)?
- • LLMs are like water, they dilute everything they touch. How do we create concentrated inputs. Dilute orange juice = gross. Dilute orange juice concentrate = good enough.
- • Building an autonomous business (or atleast a business of 1)?
- • How do we build differentiated products when the ability to build approaches zero?