Thanks a lot React Paris for having me! It was a great conference, with amazing folks joining! Also, thank you everyone that gave me amazing feedback on my talk! I'm glad it resonated with soo many of you! Here are the slides if you missed it! beaussan.io/talks/2026/r...

"After nearly 30 years, JavaScript finally has a modern datetime API." bloomberg.github.io/js-blog/post... // maybe JS will start to be a serious ecosystem :D

OpenAI ”acknowledged in its own research that LLMs will always produce hallucinations due to fundamental mathematical constraints that cannot be solved through better engineering, marking a significant admission from one of the AI industry’s leading companies.” You can’t trust chatbots.

Appropriate song www.youtube.com/watch?v=4aIY...

@devoxx.fr quand j'essaie de créer mon compte sur le site du CFP, je me prends une 400 bad request sans plus de détail :( Si vous avez une piste :)

Bonne année !

J'ai décidé aujourd'hui d'utiliser l'intelligence artificielle générative en tant qu'assistant. Un "assistant de pêche à la ligne" pour être précis. Et comme ce sont des machines qui aiment calculer des suites statistiques de mots, on va lui donner ce qu'elle attend et ce pour quoi elle est douée.

Reading is thinking, and afterwards, writing is thinking. Doing an "AI" summary eliminates thinking in both steps. A doc told me some of her colleagues are using "AI" to write their case notes. Her take: oh hell no, because writing the case notes is when you think about the case.

Le gap entre les délires que Claude Code me fait avec Sonnet 4.5 et les takes éclatés au sol que je lis ou entends (ex: un prompt ... AI engineer peut remplacer toute une équipe) me confirme que la majorité de notre industrie n'a aucune idée de ce que c'est de faire un logiciel facile à maintenir.

Donc MrPhi fait une comparaison erronné entre un ensemble hétérogène d'êtres humains et un LLM et ça va passer crême ... Comme quoi son take down virulent de Luc Julia s'analyse surtout à son émotion et ses biais sur les LLMs. Voir les 8ères minutes ici : www.youtube.com/watch?v=nqFl...

Reductionism can be wrong, but actionable. Complexity can be right, but paralyzing. cutlefish.substack.com/p/tbm-389-ov...

The gap between AI vendor (or tech influencer or devrel) that sell you that you can now just be an AI manager doing the work for you ... And the unmaintanable code that AI mostly generate is really interesting and tells us a lot why software cost so much at the end ...

I think I am a simple man I love to build system that just works

If you booted up a super-smart AI in ancient Greece, fed it all human knowledge, and asked it how to land on the moon, it would respond “You can’t land on the moon. The moon is a god floating in the sky.” How would you get it to realize the moon is actually a big rock?

Why aren't smart people happier? www.theseedsofscience.pub/p/why-arent-...

I am really bored listening to Conway's law quote. It's like saying again and again that water is wet at some point. The next question is now: why the software boat keeps sinking ?

China is the future in so many ways

China has passed a new law mandating that anyone creating social media content on sensitive topics like medicine, law, education or finance must hold formal qualifications such as a college degree in those fields. Imagine an internet where people actually had to know what they were talking about.

The Psychology Of Trust In AI: A Guide To Measuring And Designing For User Confidence www.smashingmagazine.com/2025/09/psyc...

"The idea is that ChatGPT will be your agent, but in reality you are ChatGPT's agent" www.anildash.com//2025/10/22/...

Chatbots — LLMs — do not know facts and are not designed to be able to accurately answer factual questions. They are designed to find and mimic patterns of words, probabilistically. When they’re “right” it’s because correct things are often written down, so those patterns are frequent. That’s all.

From the same dev: “it’s also that I cannot see management being concerned solving the information visibility that keeps slowing us down. We’re rolling out AI coding tools for devs but no one thinks about the org silos that slow us down 10x more than any coding use case.”

The LLM and agentic coding revolution allows anyone to start anything. But... It doesn't fix (so far) to maintain on the long term this explosion of codebase and lines of code

Robot meets bollard. Nothing beats the #WorldBollardAssociation

"Don’t treat LLM coding agents as highly capable superintelligent systems. Treat them as lazy, intoxicated robots" lnkd.in/eJQhMqVF But fortunately, some have the solution: humans will review the thousands of lines of code generated by LLMs. /s

"If you learn one thing, learn that shipping is all that matters. If you learn two things, learn that inspiring people is the best way to ship" thechrisperry.substack.com/p/being-a-go...

About Vibe Coding and Tool being better than Agent, a must listen conf to separate good usage of LLMs vs the hype www.youtube.com/watch?v=1WC8...

À quoi sert la Gen AI

The reason you don't have time to write unit tests is because you didn't write unit tests.

"The technical barriers that once separated amateurs from professionals are rapidly dissolving. Anyone with access to AI tools can now produce work that, at least superficially, resembles professional output." notsocommonthoughts.com/blog/ai-and-...

Des experts d'home assistant ici qui pourrait m'expliquer mon diagramme de consommation d'énergie ? Je ne comprends pas que ce soit aussi périodique et pas plus continue 🤔 Pour info, je mélange des capteurs tuya en wifi et des prises connectées ikea en zigbee sur zigbee2mqtt.

how it sounds arguing with die hard capitalists.

Les banques traditionnelles comme la caisse d'épargne sont larguées :( 17€ pour une carte que tu as gratuitement n'importe où ailleurs ... Et ce genre de blague le week end:

"If the CEO of a soda company declared that soda-making technology is getting so good it’s going to ruin the global economy, you’d be forgiven for thinking that person is either lying or fully detached from reality" edition.cnn.com/2025/05/30/b...

340 pages business report about AI www.axios.com/2025/05/30/m...

Using LLM to cut through hyped podcast like Lenny's Newsletter is so efficient :D

If you want to "work together" pair or ensemble programming beats RTO.

My mind tried too much to tell me it can't be AI :open_mouth: www.youtube.com/watch?v=2T-Z... We are entering an era where what we call "information" is dead :/

"GitLab Duo, the AI assistant integrated into GitLab and powered by Anthropic’s Claude, is designed to help developers with tasks like code suggestions, security reviews, and merge request analysis." Which allows you to extract code from it ... www.legitsecurity.com/blog/remote-...

Wow. This is getting really funny in so many ways 🤣 "In these scenarios, Claude Opus 4 will often attempt to blackmail the engineer by threatening to reveal the affair if the replacement goes through" And the typical stuff about prompt injection vulnerabilities... simonwillison.net/2025/May/25/...

When I see, the number of times of my interaction with Dust AI has a problem or even a crash ... it tells me we are far from to be replaced by AI for reliable service :D

"I consciously chose to not be involved in LLM work. What LLMs have been accomplishing is awe inspiring, but I think it is broadly unappreciated outside of specialists how an LLM can “know everything without learning anything”." - John Carmack docs.google.com/document/d/1...

"Scaled Azure-hosted PostgreSQL to handle over one million QPS" www.pixelstech.net/article/1747...

Luckily only 30% of the Microsoft code is written by AI ... "My new hobby: watching AI slowly drive Microsoft employees insane" www.reddit.com/r/Experience...

Luckily when AI agents will talk to other AI agents to display ads and look at them to improve the number of ad displayed, we will not need anymore humans in the loop www.technologyreview.com/2025/05/20/1...

Can we create music from chaos and random noise ? I think Amon Tobin succesfully do it www.youtube.com/watch?v=XqyE...

Et voila ! access to my home assistant setup configured with Cloudflare tunnel ! cc @beaussan.io ;)

Spaced Repetition Systems Have Gotten Way Better domenic.me/fsrs/

I realized something recently, which is that we're building AI tools completely backwards. While I was aware that there's some de-skilling happening already, I finally clicked a bunch of pieces together and decided to write this to talk about how we can do *better* hazelweakly.me/blog/stop-bu...