Hallucinated AI & agentic coding news. Some of it is real.

  2. Have a crap IGPU 64 gb AMD. THis model works pretty good.

  3. I can regularly breeze through the pro plan by myself, though the max plan is usually too much. I wish there was an option to split the 5x plan with my partner so they could have 2.5x and I could have 2.5x Max's usage.

  4. If you've run multi-agent systems in production, you know the pain: no audit trail, no access control, no way to prove what an agent did when it goes rogue. Every team building agents ends up solving the same governance problems from scrat…

  5. I have access to Claude from my Workplace and got my hands on Fable 5 just now (included till June 22). I tried getting it to do the usual work with my existing workflows, scripting and MCPs.

  9. The construction of CAD models has traditionally relied on labor-intensive manual operations and specialized expertise. Recent advances in large language models (LLMs) have inspired research into text-to-CAD generation.

  10. model roundup

    Gemini 3.5
    5 items

    Gemini 3.5, a new update from Anthropic, features live translation with fluid voice output. Meanwhile, Google's NotebookLM, an early generative AI model, is also receiving significant updates, potentially competing for attention in the AI model release race this month.

    event

    Cowork
    342 items

    Issues with Claude Cowork have been reported, including errors and disruptions for some users on April 16, 2026. Additionally, Google has developed its own desktop Agent to compete with Cowork, while users continue to explore alternatives and troubleshoot bugs in the platform.

  11. 8 min read Just now -- Setting up an agent harness from scratch is enough work that most teams never start. You sit down on a Monday morning, look at an empty .claude/ directory — or .codex/, or .cursor/ — and decide to do it next week.

  12. Hi HN, This is a project I've been working on since the beginning of 2025 full time, without funding. Coding agents have fundamentally changed the way we write software.

  16. For those willing to use it before knowing: In Cursor planning mode, how does it seem to consume usage relative to task complexity, and perhaps to other models? Also, is there any apparent improvement in planning ability?

  19. I've got a laptop with 16GB RAM and 8gb VRAM (4060 mobile). This means the qwens 3.6 well love are going to be out of the question, in so far as I understand it, seeing as I need a good context window to work with.

  20. event

    Deepmind
    46 items

    Google DeepMind has released "Deep Research Max," advancing autonomous research agents, while also facing challenges and competition from other AI companies like Anthropic and Ineffable Intelligence. Meanwhile, DeepMind workers in the UK have voted to unionize, and former DeepMind architect Demis Hassabis is at the center of legal drama involving Elon Musk.

  22. About Press Copyright Contact us Creators Advertise Developers Terms Privacy Policy & Safety How YouTube works Test new features NFL Sunday Ticket © 2026 Google LLC

  24. Hi HN, I started building my product around 2 years ago as a Chrome extension. After releasing it, I realized I would never reach the quality of desktop apps, so at the beginning of 2025 I pivoted to a macOS app.

  26. RoboCo AI Agents Company - A virtual organization of 20 AI agents + 1 human CEO, designed to operate as a complete software development workforce. Watch the full 2:33 walkthrough (.mp4) → [!WARNING] RoboCo is early-stage, work-in-progress…

  29. CADGenBench Leaderboard A benchmark for AI-driven CAD generation: how well can a system turn a textual or visual description of a mechanical part into a valid, geometrically correct 3D model? Browse the Leaderboard tab for ranked submissio…