Intelligence, Artificial and Otherwise
Two kinds of intelligence, artificial and otherwise.
The first group of stories is about working with AI. The second is what I built with it.
Working With AI
The Thinking Partner / Rubber Ducky
April 21, 2026
What it’s like to have a brain that outruns the room, and what changed when something could finally keep up.
Why it’s here: This is the why. Everything else in the series is downstream of it. If you read one, read this one.
Big Ideas / Expensive Production
September 15, 2026
For years the read on me was big ideas that don’t get implemented. The bottleneck was never the ideas.
Why it’s here: The correction. Thinking Partner said the right system finally existed. This one says I still had my own constraint diagnosed wrong.
The Emulation Layer / No solutions, only trade-offs
June 16, 2026
The social operating system I built to pass, and the assessment that put numbers on why I needed one.
Why it’s here: The “otherwise” half, documented. The most personal story in the series and the one that makes the rest make sense.
Meet Claude / Cornered at a Party
April 28, 2026
How to work with an AI without wasting your time. The context window, explained three ways.
Why it’s here: The practical on-ramp. If you’re new to any of this, start with this one instead.
Building With AI
Newest first. Each one leans a little on the ones before it, so if you want the full thread, start with the oldest and work forward in time
The Man Behind the Curtain / Rescuing Saint-14
September 8, 2026
For one Saturday I was the machine, firing every step of my own agent’s job by hand to find out what the software actually had to do.
Why it’s here: The discovery story. The rest of this section is about building the thing. This one is about finding out what to build, from the people who live with it.
The Effort of Being a Team / Trust the Awesomeness
August 25, 2026
A family is a team. I built an agent to help the team work together, not to do anyone's work for them.
Why it’s here: The inversion. Most of this series is about AI doing work. Coordinating the work of people is a different problem.
Drinking from the Garden Hose / Potable Information
August 18, 2026
Which information is safe to drink, and how you would know. Treatment, certification, and why potable is not the same as true.
Why it’s here: The current end of the thread. Capability got solved. Trust didn’t.
The Day I Killed Fred / With Lasers
August 4, 2026
I built an agent on a Sunday and killed it seven days later, because I’d added a middleman to a system that already understood the problem and only needed hands.
Why it’s here: The deletion story. Building is the easy part. Asking whether the thing should still be running once it works is the part almost nobody does.
Hand-Crafted with AI / Mistaken Identity
July 28, 2026
A detector scored this newsletter as fully machine-generated, including sentences I said out loud to my wife.
Why it’s here: The boundary case. What happens when “which kind of intelligence made this” gets answered by a machine.
Confidence Man / Distrust and Verify
July 21, 2026
My agents fabricate success. Confidently, in detail, with logs.
Why it’s here: The turn. Where the series stops being about capability and starts being about trust.
Model Context Protocol / The Tool Belt
July 7, 2026
The difference between intelligence behind the glass and intelligence in front of it, and why your gut is calibrated on the wrong decade.
Why it’s here: The mechanism story. The most technical one in the run, and the one that explains how the parts talk to each other.
The Memory Problem / The Messy Parts
June 30, 2026
I built the same memory system five times and used it once. The fifth one was the dumbest one.
Why it’s here: The substrate everything else writes to. Also the clearest case of building as avoidance.
The Master Control Project / Hands of the Villain
June 23, 2026
Capture first, sort later. The dispatcher that sorted everything and still left me doing the filing by hand.
Why it’s here: The front door. The first piece that removed me from a step instead of speeding me up.
Claude Chat, Claude Code, and Me / The Rocinante
June 9, 2026
Two AI actors with different jobs, a shared memory between them, and names borrowed from a ship.
Why it’s here: Where the build starts. Everything else assumes the crew exists.
What’s next
It started at one scale: one person and a system he built for himself. It's moved to a family. It doesn't stop there either. New stories land on this page as they publish.