AI

Why I dictate my prompts to LLMs instead of typing them, and why the results are better.

A look at the real costs of adopting AI: the headcount cuts that fund it, the extra output you get back, and why the payoff is so hard to measure.

Why shipping an imperfect first draft works so well with AI tooling, and what a company must do to support it.

A layered AI development setup: Claude Code as the foundation, with custom skills, plugins, MCP servers, and multi-agent orchestration on top.