Very useful article to setup a secured local agentic environment, controlled by an ADE (Zed here), using ACP with a sandboxed agent (docker agent in docker sandbox) using a local model (or any model) with llmman.

Same #Zed + @Docker Agent + ACP + #llmman setup as before, but now the agent runs inside a Docker sandbox (sbx), not on my machine. Same local experience, shell access boxed in, secrets stay out of reach. k33g.org/p/20261004-acp-sbx-… #DockerAgent #sbx #Security

Mon fils démarre son internat de pneumologie sur Lille en novembre. Je suis à la recherche d'un appartement soit en colocation pour deux internes, soit à l'achat. Merci à mon réseau de retransmettre

Plugged #Zed's agent panel into @Docker Agent over #ACP, then pointed it at #llmman: a coding agent that reads files, runs shell commands and answers me, entirely on my own machine. No cloud API involved. k33g.org/p/20261003-acp-zed-… #DockerAgent #LocalModels #Llmman

"vibe coding is awesome, vibe maintaining is horrible, and vibe maintaining someone else’s slop is a nightmare."

Very true. And the real challenge with AI is how we recreate an efficient learning loop with agents in the middle. Before, I tested, I failed, I learned, …. Now it includes agents/models learning loop which make the exercise very complex

I don't read code to make the code better. I read code to make myself better. I understand the system better by reading the code, which makes me better at creating new code.

That video is 2 years old. Two years later, we are still far from human-level AI. Sure, AI has superhuman performance in a number of tasks (particularly in mathematics and coding and answering questions with known answers). But where is my Level-5 self-driving car? (Before you ask, Tesla FSD is rated Level 2, Waymo is Level 4). Where is the self-driving car that can teach itself to drive in 20 hours of practice like any 17 year old? (Before you ask, we have billions of hours of training data and still can't successfully train a self-driving system by imitation learning) Where is my domestic robot? Where is the robot that can do what any 8 year-old child can do? Where is the robot that can learn a new task as quickly as an 8 year old? AI still has a hard time with the complexity and messiness of the physical world.

#llmman: "Run any agent on any model". It pulls models like OCI images and serves them behind one API that speaks OpenAI, Ollama and Anthropic. I tried it with #llamacpp as the runtime, on macOS and Linux, plus a Go example. k33g.org/p/20261001-llmman #LocalModels #llamacpp

L’ANSSI comme l’Agence Nationale de la Sécurité des Systèmes d’Information 😅

Today we’re introducing Gemini 4 Argon. It delivers frontier performance in complex workflows across real-world software engineering, knowledge work, and cybersecurity defense with an industry-leading 1M token output limit.

Next week Antwerp in Belgium, London the week after. I’m back on the road !!! Happy to be back on the road.

People who don’t play chess…

The stealth company I co-founded has a name: Hadronic. Java has a strong future, AI and agents are here to stay, and the best way to predict this future is to invent it. Why, and who we are hiring: emmanuelbernard.com/blog/202…

We addressed a few issues and feedback we received on @docker skills. We will deploy them this week on the different marketplaces (if reviews are fast enough ). Don’t hesitate to provide feedback in our issues tracker.

Have you already decided which sessions you're going to attend at @Devoxx? If you're still pondering, this online AI schedule creator might help you. State your interest, get a personalized schedule: devoxx-ai-schedule.cloud.run…

@ReBuild_Con isn't just me & @dhardiker. Meet the steering team. Audrey Neveu is a Staff Engineer & Java Champion with 15+ years of writing Java, the kind of person who knows exactly what a developer conference should feel like from the inside.

Today we launched Claude Sonnet 5.5! 30% faster and costs up to 30% less than Sonnet 5 for most work. Strong for well-scoped everyday tasks (fixing bugs, creating docs, slides). @RLanceMartin had models paint the same photo in code. You can see the jump:

personal update! i joined @OpenAI a year ago to rethink how we work in a world full of agents. tomorrow, we get to show you what we’ve been cooking 🍿 future is bright. dm if you have ideas or thoughts on making chat or codex better, ill build it

He discovered Rails is slow today