§ Writings004 entries
The archive
Essays and field notes on building multi-agent systems and steering clients through the hype cycle without crashing.
→Or browse by conceptthe index§ Find
or by topic:
001
Evals as contracts: a better way to know if your agent is working
I used to ship my clients trained models and documentation. Now I ship them contracts and the code that enforces them. The second delivery is worth about 4× the first.
2026.02.083 min readEvals
002
Why your AI pilot is stuck (and what to do about it)
If a pilot has been stuck for more than ninety days, the problem is never the model. It's ownership, quality definition, or narrative.
2025.11.223 min readStrategy
003
The three prompt patterns I actually use in production
Your prompt is not a conversation with the model. Your prompt is the body of an API call that happens to be written in English.
2025.07.143 min readPrompting
004
How to build your first AI agent in an afternoon
A working agent with hardcoded data is infinitely more valuable than an architectural diagram of a system that hasn't been built yet.
2025.04.072 min readAgents