La compétence IA que tout le monde apprend est déjà morte. Le créateur de Claude Code a arrêté de prompter. #ia #iadailynews Sources : Fortune — Boris Cherny gère des dizaines de milliers d’agents, 8× de code chez Anthropic (08/06/2026) : https://fortune.com/2026/06/08/anthropics-boris-cherny-creator-of-claude-code-says-there-are-days-he-manages-tens-of-thousands-of-ai-agents-at-once/ explainx.ai — origine « stop prompting, build loops » (Steinberger, 12 mots / 6,5 M vues ; citation Cherny) : https://explainx.ai/blog/anthropic-engineer-loops-prompts-ai-coding-harness-engineering-2026 The New Stack — Loop Engineering (Anthropic lead « ditched prompting, now writes loops ») : https://thenewstack.io/loop-engineering/ BigGo Finance — Theo : 4 PR fusionnées pendant la nuit, 20 000 $ d’inférence en 48h (~600 $/mois réels) : https://finance.biggo.com/news/0630550c86453c18 Tom’s Hardware — NVIDIA ENPIRE : 8 agents codeurs, robots installent des GPU seuls, ~99% : https://www.tomshardware.com/tech-industry/artificial-intelligence/nvidia-reveals-ai-robots-that-taught-themselves-to-install-gpus-into-motherboards-video-shows-robot-solve-high-precision-tasks-like-installing-gpus-all-by-itself Decrypt — Nvidia : des robots qui s’entraînent seuls via des agents codeurs : https://decrypt.co/371456/nvidia-built-robots-train-themselves-ai-coding-agents OpenAI — rachat d’Ona, agents Codex persistants (contexte écosystème) : https://openai.com/index/openai-to-acquire-ona/

Stop talking about Tonya, build the book he talks about your place. It was the 12-year-old who posted this book, which was published by a six million million views in a few days. The translation, until here is used there, it was taken to answer a question, read the answer, start again. The new gesture is the loop of the book. You get an objective, it is a way to verify that it is worth it. There you go, all the time, the logic, the test is on its own work, it starts, until it succeeds. Finally, the data is clear, you give the information to the machine, find it. It seems theoretical, Boris Cherny says that the code code is anthropic. He says so. "You are supposed to build a system that takes itself to a normal day. There are a few hundreds of people in parallel, it is programs that work alone. Some days, tens of thousands of years. The result is anthropic. 8 times more than the code written in January. The human has been opened to the concept of the use. What does it look like? T.O., a developer, tap a single sentence, a Sunday at 2.20 o'clock in the morning. Then, at 6.50 pm, he writes four programs. They are read by other people, corrugated, integrated during his sleep. The power of calculating the sleep is about 20,000 dollars, except that he does not pay them. His subscription costs 600 dollars by month, 600 dollars that produce equivalent to 20,000 dollars. That is the new inequality. And it is invisible. Two people, at the same post. One book. Yes, more or less, the other frame at hand. It appears on the TV. And that, at first, on the screen. We are at the bottom of the video, you can see a real robot with code code code. With graphics and a budget. Then, humans are out of the room. The robots in the "Raté Raté" article started and, at the end, installed a graphics card in a black card only 99% of the success of the comment. Here is what it changes for you. The competence that everyone takes to learn is to learn how to buy GPD. Come to become the Dactilo of 2026. What takes value is to know how to formulate an objective, quite precise, so that a machine verifies itself its work and to know where to stop it. The value of the one that executes to the one that orchestrates. The people who take advantage of it, have stopped writing letters. They put the book once. And they are. You have two options. Go up to this school. Or learn to pilot these letters yourself. That's what I teach in my training code code code of 0 to 100, even if you have never done it. The cards of the weekend are filled very quickly, formulae, en bio, or in the first comment. Bye.