The Problem Wasn't the Model
I spent the first three months treating AI like a calculator. Type a question, get an answer. Occasionally useful, mostly generic.
Blog
Posts grouped by category and tags, with searchable archives.
Results update as you type.
Don't wait too long to learn Claude Code
You're probably sick of being told to learn AI tools. But this isn't that post.
You want skills, not agents
The agent hype isn't about AI capability. It's about human org design, and we're copying the wrong thing.
Five Things AI Can't Replace — and Why Combining Them Is the Real Moat
Five things AI can't replace: trust, context, distribution, taste, and liability. These aren't product categories. They're structural layers that become more valuable as AI gets better, not less.
Why I'm Writing So Much
You've probably noticed I'm writing a lot. Here's why I keep going anyway.
33 workflows, five layers — what a real harness looks like from the inside
The piece I wrote on harnesses described five component layers that every production agentic system needs: orchestration and planning, an execution environment, validation and quality control, context and memory management, and logic and interaction.
The Inversion
For most of human history, the most valuable place to be was the middle.
The org chart was always a harness. Dorsey just built a better one.
Nobody designed the org chart as a management philosophy. It was infrastructure.
What is a harness? The engineering layer that makes AI agents actually work.
!The LLM is the engine. The harness is the car.
Delegation vs Automation
Unpopular opinion: building AI workflows that stop for human confirmation isn't careful design. It might just be discomfort dressed up as caution.
Dad, Yogi, Searching — Why the Order Matters
Dad. Yogi. Searching for what to build next. That's the order for a reason.
Most companies haven't decided how to use AI — mandate vs redesign
Two CEOs used AI to cut headcount dramatically. They made opposite bets to get there.