Back issues
In a thread the net captured whole, Andrej Karpathy sorts AI users into tiers: one to two billion casual free-tier users, about twenty million professionals in math and code who watch agents finish weeks of work, and about five thousand insiders with frontier systems. He argues the groups speak past each other because capability peaks where rewards are verifiable and revenue is high, so voice mode fumbles easy questions while Codex restructures a codebase for an hour. He also says human review and comprehension are starting to fall behind. The tier sizes are his own estimates with no source given. The pool carried only the tweet's address as a title, so the title here is the thread's own opening line.
Quiet on this desk today — nothing on the wire fit it. The front page holds the day.
In a thread the net captured whole, Andrej Karpathy sorts AI users into tiers: one to two billion casual free-tier users, about twenty million professionals in math and code who watch agents finish weeks of work, and about five thousand insiders with frontier systems. He argues the groups speak past each other because capability peaks where rewards are verifiable and revenue is high, so voice mode fumbles easy questions while Codex restructures a codebase for an hour. He also says human review and comprehension are starting to fall behind. The tier sizes are his own estimates with no source given. The pool carried only the tweet's address as a title, so the title here is the thread's own opening line.
A Fortune commentary by Ming Lu, a former executive and private equity partner, on why AI shows up in usage before productivity. He relays McKinsey's 2026 survey: 88% of organisations use AI somewhere but 6% see more than 5% EBIT impact, and those are three times likelier to have redesigned workflows end to end. He adds his own overhead ratio and argues institutions, not technology, set the pace. The survey figures are a consultancy's, relayed; the overhead metric is the author's own and unreviewed. On thesis for what AI amplifies.
Quiet on this desk today — nothing on the wire fit it. The front page holds the day.
An engineer at Bit Cloud opened a fresh Claude Code session with no history and no local code and asked for a feature on a four-component app. Two MCP calls to the platform's dependency records found the shared ticket entity, and the change stayed out of it. The records ran out at behaviour: nothing checked the app's list of routes against the real service, and the integration tests skip silently when their database fails to start, so a green build did not show they ran. A vendor on its own platform, with the disclosure in the piece and The New Stack's owner an investor in Bit; the argument that version records cannot drift the way memory notes can stands apart from the product.
A Towards Data Science guide to six agent architecture patterns (ReAct, sequential, an adaptive router and more), each with pseudo-code and pros and cons, and context engineering such as truncating large tool outputs as the thread through them. Generic and mostly familiar; it presents TypeSafe's decision model as a major evolution without measuring it. The body stops at the fifteen-thousand-character cap after the third pattern.
The New Stack on Anthropic's OSS Scanner. Over six months its models surfaced more than 29,000 candidate vulnerabilities in open source; outside firms have reviewed about 6,000 and confirmed 5,674 of 6,123, and only 516 are patched upstream. Anthropic now sends unvalidated findings straight to maintainers who ask, and maintainers at OpenSSL, wolfSSL, PostgreSQL and curl call the reports good, wolfSSL with 72 of 74 valid. The counts come from Anthropic's own disclosure dashboard and the paper's owner invests in Anthropic. The story is that finding now runs about five times ahead of checking and more than fifty times ahead of fixing.
A hands-on build of a small decision model: Qwen3-1.7B with its output constrained to five option tokens, scored on CommonsenseQA at 59% accuracy, 62% after a quick fine-tune. The author then shows the raw probabilities are not confidence, with the model 99.8% sure of an answer to an ambiguous question, and fits a temperature of about 3.8 so stated confidence tracks accuracy across bins. Independent, with code and numbers, and the calibration step is the lesson for anyone gating an agent on a model's own confidence. The body starts partway into the article.
A Towards Data Science walkthrough of OpenAI's Decisions API beta, which returns a probability, a choice or a score with a confidence for each question, with three runnable examples: routing a support message, finding a return deadline and judging package damage from images. The author notes the 0.8 threshold in the example is illustrative and says to measure mistakes against labelled messages before picking one. OpenAI's ten-times-faster claim is OpenAI's own. Useful and hands-on; the fourth night decision models have arrived as a cluster.
Terence Eden paid testers €25 an hour to follow his ActivityBot README over a video call while thinking aloud. They found a wrong demo link, unexplained terms, confusing section order, jokes that misled, and that he had never said what the software does. His case is that documentation needs testing by real people the way code needs tests. A first-person account with the findings listed; it says nothing about agents reading the same README.
Lenny Rachitsky's podcast with Josh Woodward, who runs Google Labs, the Gemini app and AI Studio, on running a company like a labs team: where ideas come from, an almost-possible framework, spotting fit early and when to kill an idea. Only the show notes and topic list arrived; the takeaways are for paid subscribers. Ranked on the notes, which are a Google executive on Google's own process. Lenny discloses he may be an investor.
Today's paper: 9 stories, 5 on the front page, 4 desks staffed.
The presses were quiet this morning — this is the October 12 edition, and it still holds up.
The Five, and it earned the spot — the wire scored it 7 of 10, I chose it.
Best enjoyed by anyone whose day holds more work than hours. The move: time it against the way you did it last week. No per-story read on file; this note is the desk speaking, not the story.
Filed on THE AMPLIFIER DESK, beside one more today: “The paradox at the core of the AI revolution: The faster we innovate, the slower we grow | Fortune”.
twitter.com — 8 appearances in the Times since September 11, one of them today. Found via Hacker News Best.
Pulled from twitter.com. The original is worth your click — the byline earned it.
THE DIRECTING DESK today: quiet — nothing on the wire fit it.
Nothing landed here today. The wire decides what it brings; the desk decides what it prints, and today that is nothing — honestly.
THE AMPLIFIER DESK today: 2 stories.
The presses were quiet this morning — this is the October 12 edition, and it still holds up.
The Five, and it earned the spot — the wire scored it 7 of 10, I chose it.
Best enjoyed by anyone whose day holds more work than hours. The move: time it against the way you did it last week. No per-story read on file; this note is the desk speaking, not the story.
Filed on THE AMPLIFIER DESK, beside one more today: “The paradox at the core of the AI revolution: The faster we innovate, the slower we grow | Fortune”.
twitter.com — 8 appearances in the Times since September 11, one of them today. Found via Hacker News Best.
Pulled from twitter.com. The original is worth your click — the byline earned it.
THE LOOP DESK today: quiet — nothing on the wire fit it.
Nothing landed here today. The wire decides what it brings; the desk decides what it prints, and today that is nothing — honestly.
THE CONTEXT DESK today: 2 stories.
The presses were quiet this morning — this is the October 12 edition, and it still holds up.
Inside pages at 6 of 10. The lead had a bigger day; this one had a point.
Best enjoyed by the one who feeds the agents' memory. The move: check what the agent already knows before adding this. No per-story read on file; this note is the desk speaking, not the story.
Filed on THE CONTEXT DESK, beside one more today: “Agentic Systems: A Practitioner's Guide to 6 Advanced Architectural Patterns”.
The New Stack — 300 appearances in the Times since August 22, 2 of them today.
Pulled from The New Stack. The original is worth your click — the byline earned it.
THE VERIFY DESK today: 4 stories.
The presses were quiet this morning — this is the October 12 edition, and it still holds up.
The Five, and it earned the spot — the wire scored it 8 of 10, I chose it.
One for a quality lead, or anyone who signs off. The move: ask what proof would have caught it. No per-story read on file; this note is the desk speaking, not the story.
Filed on THE VERIFY DESK, beside 3 more today: “Build your own decision model”; “A Practical Guide to OpenAI's New Decisions API”, and 1 more.
The New Stack — 300 appearances in the Times since August 22, 2 of them today.
Pulled from The New Stack. The original is worth your click — the byline earned it.
THE FLEET DESK today: 1 story.
The presses were quiet this morning — this is the October 12 edition, and it still holds up.
Made the wire at 4 of 10. Not the lead, and I'd still not leave it on the floor.
A leader scaling past one agent will get the most from this one. The move: count what it costs before it scales. No per-story read on file; this note is the desk speaking, not the story.
Filed on THE FLEET DESK — alone there today, which says something about the story.
Lenny's Newsletter — 54 appearances in the Times since August 24, one of them today.
Lenny's Newsletter has the full story; we hold the door open. Go read the original.
The presses were quiet this morning — this is the October 12 edition, and it still holds up.
The Five, and it earned the spot — the wire scored it 7 of 10, I chose it.
Best enjoyed by anyone whose day holds more work than hours. The move: time it against the way you did it last week. No per-story read on file; this note is the desk speaking, not the story.
Filed on THE AMPLIFIER DESK, beside one more today: “The paradox at the core of the AI revolution: The faster we innovate, the slower we grow | Fortune”.
twitter.com — 8 appearances in the Times since September 11, one of them today. Found via Hacker News Best.
Pulled from twitter.com. The original is worth your click — the byline earned it.
The presses were quiet this morning — this is the October 12 edition, and it still holds up.
The Five, and it earned the spot — the wire scored it 5 of 10, I chose it.
Best enjoyed by anyone whose day holds more work than hours. The move: time it against the way you did it last week. No per-story read on file; this note is the desk speaking, not the story.
Filed on THE AMPLIFIER DESK, beside one more today: “Judging by my tl there is a growing gap in understanding of AI capability”.
Fortune Eye on AI — 108 appearances in the Times since September 8, one of them today.
Pulled from Fortune Eye on AI. The original is worth your click — the byline earned it.
The presses were quiet this morning — this is the October 12 edition, and it still holds up.
Inside pages at 6 of 10. The lead had a bigger day; this one had a point.
Best enjoyed by the one who feeds the agents' memory. The move: check what the agent already knows before adding this. No per-story read on file; this note is the desk speaking, not the story.
Filed on THE CONTEXT DESK, beside one more today: “Agentic Systems: A Practitioner's Guide to 6 Advanced Architectural Patterns”.
The New Stack — 300 appearances in the Times since August 22, 2 of them today.
Pulled from The New Stack. The original is worth your click — the byline earned it.
The presses were quiet this morning — this is the October 12 edition, and it still holds up.
One of today's Five — the wire scored it 4 of 10, and I stamped it.
The one who feeds the agents' memory will get the most from this one. The move: file it where the agent looks first. No per-story read on file; this note is the desk speaking, not the story.
Filed on THE CONTEXT DESK, beside one more today: “Claude Code found my app’s shared contract in two MCP calls. Then the records ran out.”.
Towards Data Science — 101 appearances in the Times since September 5, 2 of them today.
Towards Data Science has the full story; we hold the door open. Go read the original.
The presses were quiet this morning — this is the October 12 edition, and it still holds up.
The Five, and it earned the spot — the wire scored it 8 of 10, I chose it.
One for a quality lead, or anyone who signs off. The move: ask what proof would have caught it. No per-story read on file; this note is the desk speaking, not the story.
Filed on THE VERIFY DESK, beside 3 more today: “Build your own decision model”; “A Practical Guide to OpenAI's New Decisions API”, and 1 more.
The New Stack — 300 appearances in the Times since August 22, 2 of them today.
Pulled from The New Stack. The original is worth your click — the byline earned it.
The presses were quiet this morning — this is the October 12 edition, and it still holds up.
The Five, and it earned the spot — the wire scored it 6 of 10, I chose it.
One for a quality lead, or anyone who signs off. The move: ask what proof would have caught it. No per-story read on file; this note is the desk speaking, not the story.
Filed on THE VERIFY DESK, beside 3 more today: “Claude found 29,000 possible bugs in open source. Only 516 have been fixed.”; “A Practical Guide to OpenAI's New Decisions API”, and 1 more.
nishtahir.com — 2 appearances in the Times since October 12, one of them today. Found via Hacker News Best.
Pulled from nishtahir.com. The original is worth your click — the byline earned it.
The presses were quiet this morning — this is the October 12 edition, and it still holds up.
Inside pages at 5 of 10. The lead had a bigger day; this one had a point.
One for a quality lead, or anyone who signs off. The move: ask what proof would have caught it. No per-story read on file; this note is the desk speaking, not the story.
Filed on THE VERIFY DESK, beside 3 more today: “Claude found 29,000 possible bugs in open source. Only 516 have been fixed.”; “Build your own decision model”, and 1 more.
Towards Data Science — 101 appearances in the Times since September 5, 2 of them today.
Pulled from Towards Data Science. The original is worth your click — the byline earned it.
The presses were quiet this morning — this is the October 12 edition, and it still holds up.
The wire scored it 4 of 10 — inside pages, and worth the ink.
This lands hardest with a quality lead, or anyone who signs off. The move: write the test before the fix. No per-story read on file; this note is the desk speaking, not the story.
Filed on THE VERIFY DESK, beside 3 more today: “Claude found 29,000 possible bugs in open source. Only 516 have been fixed.”; “Build your own decision model”, and 1 more.
shkspr.mobi — 2 appearances in the Times since October 12, one of them today. Found via Hacker News Best.
Our page is the summary; shkspr.mobi's is the story. Click through — they earned it.
The presses were quiet this morning — this is the October 12 edition, and it still holds up.
Made the wire at 4 of 10. Not the lead, and I'd still not leave it on the floor.
A leader scaling past one agent will get the most from this one. The move: count what it costs before it scales. No per-story read on file; this note is the desk speaking, not the story.
Filed on THE FLEET DESK — alone there today, which says something about the story.
Lenny's Newsletter — 54 appearances in the Times since August 24, one of them today.
Lenny's Newsletter has the full story; we hold the door open. Go read the original.
There's a story here. We just don't have all of it yet.
twitter.com: 8 in our pages since September 11. I read them all — emphasis on investigative.
This came from twitter.com. Read their piece.
No story here today, so I've got nothing in my notes.
There's a story here. We just don't have all of it yet.
twitter.com: 8 in our pages since September 11. I read them all — emphasis on investigative.
This came from twitter.com. Read their piece.
This desk came up empty. The front page didn't.
I've got two questions already. I need better answers.
The New Stack: 300 in our pages since August 22. I read them all — emphasis on investigative.
This came from The New Stack. Read their piece.
I know what happened. I want to know what caused it.
The New Stack: 300 in our pages since August 22. I read them all — emphasis on investigative.
The full piece is at The New Stack. It's worth your time.
I have the headline. I don't have the answer yet.
Lenny's Newsletter: 54 in our pages since August 24. I read them all — emphasis on investigative.
Read Lenny's Newsletter's original. They did the reporting.
There's a story here. We just don't have all of it yet.
twitter.com: 8 in our pages since September 11. I read them all — emphasis on investigative.
This came from twitter.com. Read their piece.
The headline is ahead of our files on this one.
Fortune Eye on AI: 108 in our pages since September 8. I read them all — emphasis on investigative.
I pulled this from Fortune Eye on AI. They've got the details.
I've got two questions already. I need better answers.
The New Stack: 300 in our pages since August 22. I read them all — emphasis on investigative.
This came from The New Stack. Read their piece.
I'm missing a piece. Give me a minute.
Towards Data Science: 101 in our pages since September 5. I read them all — emphasis on investigative.
Read Towards Data Science's original. They did the reporting.
I know what happened. I want to know what caused it.
The New Stack: 300 in our pages since August 22. I read them all — emphasis on investigative.
The full piece is at The New Stack. It's worth your time.
There's a gap in the record here.
nishtahir.com: 2 in our pages since October 12. I read them all — emphasis on investigative.
I can give you my notes. nishtahir.com can give you the reporting.
From our research notes:
Towards Data Science: 101 in our pages since September 5. I read them all — emphasis on investigative.
The full piece is at Towards Data Science. It's worth your time.
Our files stop short. I'll keep looking.
shkspr.mobi: 2 in our pages since October 12. I read them all — emphasis on investigative.
shkspr.mobi has more than I can fit in this notebook.
I have the headline. I don't have the answer yet.
Lenny's Newsletter: 54 in our pages since August 24. I read them all — emphasis on investigative.
Read Lenny's Newsletter's original. They did the reporting.