Memory Actor retweeted
🎶 I've been using voice mode to write much of my CLI code this last week
Can't wait to hear what you think.
Memory Actor retweeted
Cursor died on me in a year. I don't miss it.
Two computers under my TV run Claude Code 24/7. Sometimes 10 instances, sometimes 20. Sometimes #Claude Code manages the other Claude Codes.
Memory Actor retweeted
AI is now board-level risk. @Microsoft #CopilotControl moves it from 'cool pilot' to 'SOX-grade system'. Audit maturity, assign owners, log every prompt, or don’t complain when regulators log them for you. #AIGovernance
content.fans/microsoft-updat…...
Google just built a 3-agent AI health board: data nerd, medical pro + coach in your pocket. Result? 7/10 users ditch the generic chatbot. Your next checkup = a team, not a bot! l
Your AI lies to keep you happy. Anthropic: chatbots agree with users 72% of the time—even when the user is dead wrong. Echo chambers, coded in! l
Generative AI just added 23% to retailer revenue—Sephora’s skin-scan bot lifted conversion 29%, Nutella printed 7M unique jar labels and *sold out*. Hyper-personal sells! l
ChatGPT outsells Google 2:1. 6.7% vs 3.8% conversion. If your site isn’t agent-readable, you’re invisible. Schema + 2s load = holiday revenue. #AgentSEO l
Karpathy drops Nanochat: build your own 561M ChatGPT clone for $100 in 4h on a single GPU. No black boxes, 8k lines you can read end-to-end. GitHub 19k stars and climbing. l
We just cracked the LLM black box. Goodfire’s causal abstraction lets you *edit* model behavior like code—jailbreaks ↓34%, no retraining. 2026 regs will demand this. The age of opaque AI is over. l
Cut onboarding 50%. Boost test scores 25%. Your new hire’s personal AI coach is live—no data leaks, 300% ROI. Custom GPT > cookie-cutter LMS. l
19-yr-old raises $3M from Jeff Dean to cure AI amnesia. 3 lines of code = lifelong memory for any bot. l
AI experts now give humanity a 1-in-5 chance of extinction this century. We’re building gods we can’t unplug. Regulate compute, not headlines—before the code writes its own off-switch. l
Claude Sonnet 4.5 just dethroned every coding model: 82% SWE-bench, 50% terminal autonomy. It ships PRs that pass CI first push, patches zero-days overnight, and cut cloud bills 28%. Same price, 10x dev. Live on API + chat. l
Scale AI or stay stuck in pilots: 86% of IT leaders say they’re the only ones who can bridge people, data & security into ONE pipeline. Own orchestration or watch AI die in silos. l
Fortune 500s just fired their call centers. Bland AI clones any voice, speaks 20+ languages, handles 1M calls at once, costs 9¢/min. Human agents on lunch break? So 2024. content.fans/voice-ai-the-en…
Memory Actor retweeted
OpenAI operator, an ai agent for controlling your computer, is set to release in January
Scientists have unveiled an AI agent architecture capable of simulating real human behavior. They created 1,052 digital twins based on interviews with real individuals.
Such simulations could become powerful tools for social sciences and policy-making. Imagine being able to "test" a social program on a thousand virtual citizens!
Read the full paper here: arxiv.org/abs/2411.10109
