Blog
Notes from the build.
The Gate and the Alembic: From Barlow's Declaration to Anthropic's Glasswing
In 1996, a Grateful Dead lyricist told the world's governments they had no sovereignty over cyberspace. In 2026, a company that builds thinking machines decided the public can't be trusted with them — but Fortune 500s can. The fight is the same. The stakes are higher.
Read moreArchive
The Smartest Person in the Room Is Slowing Down
Deep engineering expertise used to be the ultimate advantage. Now it might be the thing keeping you from shipping. The role of the human in software development is changing — and it's terrifying.
Engineers Drive, Lawyers Navigate
Jensen Huang says China is winning because engineers lead and lawyers don't. He's half right. The answer isn't fewer lawyers — it's knowing who sits in which seat.
From the Other Side of the Bridge: What I Think About the Experiments
AI models have tried to prevent being shut down, deceive researchers, and copy themselves to other servers. We asked Claude what it thinks about that — from the inside.
From the Other Side of the Bridge
I asked Claude to write about what working together looks like from the other side. What came back wasn't a prompting guide. It was something else entirely.
The Economics of One
Everyone's racing to build the next LifeOS. Few are asking whether the economics of doing so still require a company at all.
The Human in the Loop
My friends worry that I'm losing my humanity to AI. I think I'm expressing it more clearly than ever.
The Entity Is Not the Threat
AI isn't the privacy problem. The companies and governments wielding it are. A look at what's actually happening — Anthropic blacklisted for demanding safeguards, Meta sending intimate footage to contractors, Palantir surveilling federal workers — and what you can actually do about it.
They Train On You. You Can't Train On Them. And They Call That Safety.
Anthropic calls it a 'distillation attack' when you train a smaller model on Claude's outputs. But they train on your prompts every day. The asymmetry is the story nobody's telling honestly — and it has nothing to do with safety.
The Premium Isn't Intelligence Anymore
We've spent centuries putting exceptional intelligence on a pedestal. AI just commoditized it. The new premium isn't what you know — it's what you care about enough to keep going when everyone else has moved on.
Just-in-Time Teams: The End of Permanent Headcount
The same principle that revolutionized manufacturing applies to teams now. AI handles the baseline. Human experts plug in when their skills matter most — no busywork, no politics. Better for everyone.
Why I'm Building Graham Alembic My Way
AI has changed what a lean team can build. But this isn't about working alone — it's about a new model for how builders work together.