Blog posts
Thoughts on AI, technical writing, and software.
Claude Code beyond prompting
Turns out I'd been using a Swiss Army knife with only the main blade
Code execution with MCP: separating context from computation
TL;DR: Traditional tool calling eats away at an agent's context window. All the tool definitions and every intermediate result pass through it.
On agency, abstraction, and AI-assisted coding
I still feel like I'm using my brain?
Testing docs IA with AI agents
Testing information architecture with agents, impatience, and one very frustrated bot.
Can AI help test IA? A Wikipedia game experiment
I don't actually know, but it reminded me of a few things about human navigation.
The Exploratorium boulder: why enthusiasm isn't always contagious
On becoming the colleague I used to roll my eyes at.
Docs metrics and the stories we tell ourselves
My roles prior to developer education were in site reliability and infrastructure.
How onboarding a human made my AI smarter
Apparently neither humans nor AI can read my mind?
Digging into "Your Brain on ChatGPT"
How educational is a 20 minute SAT essay, really?
Learn MCP with me, part 2: working with MCP servers and clients
Okay, I'm actually excited about MCP now.
Learn MCP with me, part 1: what is this even for?
Because LLMs can't stop hallucinating parameter names.
We all fall down: notes from an ordinary outage
On loops, priorities, and the fragility of complex systems.