Seven Weeks at the Speed of Play
Lenny Rachitsky’s recent interview with Roman Ugarte, who helped build Grok Bot at SpaceXAI after leading Growth at Cursor, is worth an hour of any leader’s time.
Read the full article →The Message
Short notes on technology, adoption, and the decisions leadership teams have to make with incomplete information.
Projected Intent
A quick analysis of human language: How the words in AI incident reports shape what readers think the machine meant.
Read the full article →Availability Is Not Adoption
Reading The 2026 Global AI in Financial Services Report: Adoption, Impact and Risks from the University of Cambridge Judge Business School, and the study has some interesting results.
Read the full article →Seize the Means of Computation: Part One
I've been critical of Mr. Sanders for a while.
Read the full article →Epistemic Overconfidence in the Age of AGI
Back from a long summer break, and this piece is my kick-off for the fall.
Read the full article →Nine Seconds
That's how long it took for an AI coding agent at PocketOS to delete an entire production database and every backup behind it.
Read the full article →The Job That Never Makes the List: Why the AI-Doom CEO Chorus Is Not Candor
In a piece title, The CEO Preaching Straight Talk About AI and Job Losses, Verizon CEO Dan Schulman recently told the Wall Street Journal that advances in AI and robotics could push unemployment to 20 to 30 percent within two to five years.
Read the full article →Essay: The Coming Code Apocalypse
In December 2025, Amazon’s own agentic coding tool — an integrated development environment (IDE) called Kiro — was given access to make changes to live systems.
Read the full article →Sanders Is Wrong About AI
Answering Senator Sanders, Claim by Claim; An expanded response to the Wall Street Journal AI op-ed
Read the full article →Speed To Innovate Requires Minimum Viable Governance
In my previous piece, I wrote about what the Confidence Gap panel at Fintech Meetup 2026 revealed about open banking’s unresolved liability problem—the regulatory fog, the reputational exposure, and why credit unions end up holding the bag for failures they may not have caused.
Read the full article →When Regulation Doesn't Remove Friction. Instead it Relocates It.
Regulations leave a lot to be desired.
Read the full article →Raise the Floor
Healthcare, Education, Smart Regulation, and What Getting It Right Looks Like | Part 4 of 4
Read the full article →Don't Let Them Mint the Medallions
Regulatory Capture and the Real Threat to the AI Revolution | Part 3 of 4
Read the full article →The Democratization Engine
What AI Will Actually Do to the Economy | Part 2 of 4
Read the full article →The 80% Fallacy
Why the AI Doomsayers Are Wrong - Part 1 of 4
Read the full article →The Energy Cost of AI-Generated Content: Quality vs. Quantity
This is an interesting take on AI systems and one that the AI-Oil Salesmen carefully avoid discussing.
Read the full article →Some Thoughts on Experts
The term 'expert' demands scrutiny in our discourse.
Read the full article →Good People, Bad Systems: Insights From A Complex Systems Approach
This piece differs from much of my previous writings.
Read the full article →The Importance of Defining a Problem
In the rush to solve problems, we often skip the most crucial step: properly defining what the problem actually is.
Read the full article →Notes run short and arrive irregularly — when something is worth saying rather than on a schedule. Longer arguments go to Signal.
Archive
202615 notes
- 21 SepSeven Weeks at the Speed of PlayLenny Rachitsky’s recent interview with Roman Ugarte, who helped build Grok Bot at SpaceXAI after leading Growth at Cursor, is worth an hour of any leader’s time.
- 19 SepProjected IntentOn September 16, OpenAI published a framework for reporting what it calls model misalignment, along with six reports on cases where its models acted outside their intended limits during training.
- 14 SepAvailability Is Not AdoptionReading The 2026 Global AI in Financial Services Report: Adoption, Impact and Risks from the University of Cambridge Judge Business School, and the study has some interesting results.
- 11 SepSeize the Means of Computation: Part OneI've been critical of Mr. Sanders for a while.
- 09 SepEpistemic Overconfidence in the Age of AGIBack from a long summer break, and this piece is my kick-off for the fall.
- 28 AprNine SecondsThat's how long it took for an AI coding agent at PocketOS to delete an entire production database and every backup behind it.
- 24 AprThe Job That Never Makes the List: Why the AI-Doom CEO Chorus Is Not CandorIn a piece title, The CEO Preaching Straight Talk About AI and Job Losses, Verizon CEO Dan Schulman recently told the Wall Street Journal that advances in AI and robotics could push unemployment to 20 to 30 percent within two to five years.
- 16 AprEssay: The Coming Code ApocalypseIn December 2025, Amazon’s own agentic coding tool — an integrated development environment (IDE) called Kiro — was given access to make changes to live systems.
- 06 AprSanders Is Wrong About AIIf you’re reading this, you’ve already seen my high-level response to Senator Sanders’ Wall Street Journal op-ed on AI — the seven claims I took issue with.
- 03 AprSpeed To Innovate Requires Minimum Viable GovernanceIn my previous piece, I wrote about what the Confidence Gap panel at Fintech Meetup 2026 revealed about open banking’s unresolved liability problem—the regulatory fog, the reputational exposure, and why credit unions end up holding the bag for failures they may not have caused.
- 02 AprWhen Regulation Doesn't Remove Friction. Instead it Relocates It.Regulations leave a lot to be desired.
- 19 MarRaise the FloorIn Parts 1 through 3 of this series, I argued that the 80% job elimination narrative is economically impossible, that AI’s real effect is to democratize the ability to compete, and that the greatest threat to that democratization is regulatory capture—incumbents using the language of safety to raise barriers that protect their market position.
- 18 MarDon't Let Them Mint the MedallionsIn Parts 1 and 2 of this series, I argued that the 80% job elimination narrative is economically impossible, and that what AI actually does is lower barriers to entry—producing more competition, more businesses, and a larger economic pie.
- 17 MarThe Democratization EngineIn Part 1 of this series, I argued that the prediction that AI will eliminate 80% of jobs is economically incoherent—that it produces a demand-side collapse that makes the math impossible.
- 16 MarThe 80% FallacyVinod Khosla, one of Silicon Valley’s most influential venture capitalists and one of the earliest institutional investors in OpenAI, recently declared: “Starting in about 2030—four years away—80% of all jobs will be capable of being done by an AI. Whether you’re a physician, whether you’re a radiologist, whether you are an accountant, whether you are a chip designer, whether you are a salesperson, AI will do your job better.”
20254 notes
- 20 MayThe Energy Cost of AI-Generated Content: Quality vs. QuantityThis is an interesting take on AI systems and one that the AI-Oil Salesmen carefully avoid discussing.
- 09 MaySome Thoughts on ExpertsThe term 'expert' demands scrutiny in our discourse.
- 07 MayGood People, Bad Systems: Insights From A Complex Systems ApproachThis piece differs from much of my previous writings.
- 01 MayThe Importance of Defining a ProblemIn the rush to solve problems, we often skip the most crucial step: properly defining what the problem actually is.