How To Be A World-Class Agentic Engineer
The only framework you need. Strip the dependencies. Own the context. Let the foundation companies do the rest.
Intelligence Brief
Deep dives, syntheses, and weekly briefs on the forces reshaping education, AI, and the world economy.
What if AI bullishness is right — and that's actually bearish? Ghost GDP, the SaaS collapse, and the intermediation wipeout. The scenario nobody wants to read but everyone should.
In 2022, 800,000 human neurons learned to play Pong. Not metaphorically. Here is what that means — and why it matters more than almost anything happening in AI right now.
Daniel Kokotajlo mapped out what happens if AI scaling doesn't break. It reads like fiction. The math suggests it probably isn't.
The only framework you need. Strip the dependencies. Own the context. Let the foundation companies do the rest.
Most AI agents don't fail because of the model. They fail because of the context. And most people have no idea they're the ones doing the poisoning.
70 million white-collar workers. 12 to 18 months. Andrew Yang calls it the great disemboweling — here's what the divide looks like from the inside.
You know more about AI than 99% of people. You've read the threads, watched the demos. And you've built almost nothing with it. That's the trap.
Prestige used to be "I take risks." Now it's "I can't lose and I still might win big." How the dream changed — and where to place your bet.
AI isn't just automating tasks. It's exposing how much of our necessary work was always a story. The job religion is losing its monopoly.
Anthropic opened Claude and found something that looks like a mind. Not by design — the algorithm just produces this when you give it enough room to run.
Most skills will be irrelevant in 10-20 years. One won't. And half the population has already been trained out of it.
Info products aren't dead. The format is. Here's what the next-generation one-person business looks like in 2026 — and why the window is only 2-3 years wide.
Dario Amodei confronts the real risks of powerful AI. Not doom. Not denial. A surgical examination written by the person building it — including a warning about Anthropic itself.
52% of graduates are underemployed. January 2026 was the worst layoff month in 17 years — while the economy grew at 4%. The deal made with a generation is dead.
Anthropic vs. the Pentagon. A student deported mid-semester. OpenAI's insider trading scandal. Seven stories that define the moment.
Tim Denning's counter-narrative to doom. Not collapse — a forced upgrade. High agency is the new baseline. And the window is shorter than you think.
The Cursor CEO: 35% of their PRs are now created by autonomous cloud agents. Tab is dead. Synchronous agents are fading. The fleet era is here.
Ghost GDP, the Friction concept, the doom loop, the cascade. Everything in the report that crashed the stock market — explained.
Bloomberg costs $2,000/month. Claude costs $20. Here are the 10 prompts that close the gap — plus the real capital opportunities hiding in plain sight right now.
Quantum encryption cracked at 100K qubits. Brain cells play Doom. Waymo hits 10 cities. Meta's AI Safety head deletes her own inbox. 11 stories, no fluff.
Folder structure, context files, prompt formula, power moves — everything you need to walk away and come back to finished work. 1.9M views.
The only framework you need. Strip the dependencies. Own the context. Let the foundation companies do the rest.
Andrew Yang calls it 'The Fuckening' — the great disemboweling of white-collar jobs. 70 million workers. 12 to 18 months. Here's what it actually means.
AI isn't just automating tasks. It's exposing how much of our so-called necessary work was always a story. The job religion is losing its monopoly.
Most skills will be irrelevant in 10-20 years. One skill won't. The case for agency — and why half the population has already opted out.
You know more about AI than 99% of people. You've read the threads, watched the demos, followed the launches. And you've built almost nothing with it. That's the trap.
Anthropic opened Claude and found something that looks like a mind. Not because they designed it that way. Because the algorithm just produces this when you give it enough room to run.
Most AI agents don't fail because of the model. They fail because of the context. And most people have no idea they're the ones poisoning it.
Info products aren't dead. The format is. Here's what the next-generation one-person business looks like in 2026 — and why the window is only 2-3 years wide.
Prestige used to be 'I take risks.' Now it's 'I can't lose and I still might win big.' How the dream changed — and what it means for where you place your bet.
Next brief dropping soon
Daily at 8AM — or drop a link