Sponsored by
🧠 THAT ONE AI - This Week’s Signal
Here’s what’s shaping AI right now - without the noise:
🖥️ Claude can now run your computer in the background.
💰 Nvidia buys Hugging Face for $12.9 billion.
🧠 Claude Fable 5.1 Just Reset the Coding Benchmark
📒 That One AI Tip: setting up background computer use without getting burned.
🧰 Tools worth testing
🖥️ Claude Can Now Run Your Computer While You Do Something Else
Anthropic shipped background computer use for Claude Cowork and Claude Code. Claude now operates your desktop directly, clicking and typing through apps without you watching every step, working in background windows on macOS 15 or later while you keep using your machine for other things.
It's in beta on Pro and Max plans, and only reaches apps you've explicitly approved. Claude asks permission before touching each new app, and some categories, like trading and investment platforms, are blocked by default. Anthropic also published a prompting guide built specifically for Fable 5.1, aimed at getting the model to write more clearly.
Turning it on takes one settings toggle: Settings → General → Enable computer use.
The bigger signal: 👉 Background computer use closes the gap between AI that helps you work and AI that works while you don't. That gap has been the whole point of agentic AI, and it just got a lot smaller.
A Free Guide From HubSpot How to Turn AI into Your Income Engine
Ready to transform artificial intelligence from a buzzword into your personal revenue generator? Our groundbreaking guide "200+ AI-Powered Income Ideas" is your gateway to financial innovation in the digital age.
Inside you'll discover:
• A curated collection of 200+ profitable opportunities spanning content creation, e-commerce, gaming, and emerging digital markets—each vetted for real-world potential
• Step-by-step implementation guides designed for beginners, making AI accessible regardless of your technical background
• Cutting-edge strategies aligned with current market trends, ensuring your ventures stay ahead of the curve
Download your guide today and unlock a future where artificial intelligence powers your success. Your next income stream is waiting.
💰 Nvidia Just Bought Hugging Face for $12.9 Billion
Nvidia agreed to acquire Hugging Face, the open-source platform millions of developers use to share models and datasets, for $12.9 billion. It's Nvidia's second-biggest purchase ever, behind only the $20 billion Groq asset deal late last year.
Jensen Huang says Hugging Face will stay "an open platform for the entire AI ecosystem." Clément Delangue, Hugging Face's co-founder and CEO, framed the deal as proof that open-source AI has become a real complement, and sometimes an alternative, to closed frontier labs, ten years after he started the company.
The deal is expected to close in the first half of next year, pending regulatory approval.
The bigger signal: 👉 Nvidia now owns the hardware most models train on and the repository most developers share models through. If you build on open-source AI, your stack now touches one company at two different layers.
🧠 Claude Fable 5.1 Just Reset the Coding Benchmark
Anthropic shipped Claude Fable 5.1 and Claude Mythos 5.1 this week, and the benchmark jump is the real story. Fable 5.1 scored 55.8% on Terminal-Bench 4.0, up from 42.0% for Fable 5, and more than doubled the previous score on Terminal-Bench-Science, hitting 52.6%.
Fable 5.1 also gets cheaper as it gets smarter. Cache reads cost 75% less than Fable 5's, cutting real-world costs around 25% for typical work and up to 45% on heavy agentic tasks. Anthropic paired the release with Enterprise Frontier Safeguards, a zero-data-retention privacy layer rolling out in phases starting this fall.
Mythos 5.1, the sibling model built for cyberdefenders and life scientists, isn't going to the general public. It's available only through Anthropic's trusted access programs for screened researchers, with fewer guardrails than the consumer-facing model.
The bigger signal: 👉 Anthropic is splitting its model lineup: a public version with full guardrails, and a restricted version with fewer of them for vetted researchers only. Safety and capability don't have to move at the same speed for everyone.
📒 That One AI Tip: Setting Up Claude's Background Computer Use Without Getting Burned
Claude can now operate your desktop directly while you do something else, in both Cowork and Claude Code. Genuinely useful. Also the kind of feature worth setting up carefully before you hand it real work.
Turn it on deliberately, not by default. Go to Settings → General, under Desktop app, and flip Enable computer use. It's opt-in on purpose: once it's live, Claude can reach any app you approve, so don't flip the switch and walk away without reading the rest of this.
Block the apps that actually matter yourself. Claude is trained to avoid banking, healthcare, government, and legal sites, and blocks trading and investment platforms by default. Trained isn't the same as guaranteed. Add anything with real financial or medical stakes to the app blocklist instead of trusting the guardrail to hold on its own.
Start small before you start complex. Research and file organizing are where computer use is genuinely reliable right now. Multi-step workflows across several apps are still where it trips. Build trust on the easy stuff first.
Close what you don't want it to see. Claude screenshots your screen to navigate it, so anything visible, including apps you didn't explicitly grant access to, is fair game if it's on screen. Close sensitive tabs and documents before you start a session.
The bigger signal: 👉 The permission model matters more than the feature itself. Claude asking before every new app is what separates useful automation from automation you can't audit. Don't turn that friction off just to save a few clicks.
That One AI 🧰 TOOLBOX
A few tools quietly worth exploring:
⚙️ Muse Code → Meta's terminal coding agent, now out of beta, built to coordinate multiple agents on a single task.
🎬 CinLink → Creates, edits, translates, dubs, and repurposes video and audio in one pass, useful for anyone shipping content across languages
📞 ThunderPhone → Builds AI phone agents starting at 2 cents a minute across 47 languages, no telephony infrastructure required.
🔧 Edge Drop → Open-source Windows clipboard shelf that lives on your screen edge, letting you stack and drag text or files into any app.
The best voice models, now across all channels
Most CX platforms do not own the voice. They orchestrate a workflow, then call a third party for speech and transcription. Every hop adds latency, cost, and another vendor to manage.
ElevenAgents is the opposite. They make the voice models the market builds on, and ElevenAgents puts full orchestration on top. Voice, transcription, text-based chat, and reasoning run in one vertically integrated pipeline, so responses come back in <400 milliseconds and sound human, not synthetic.
Plus, you keep full control. Plug in any LLM, integrate tools, webhooks, and MCP servers, and ground responses in your knowledge base. Get an agent live in minutes, then A/B test with Experiments, enforce Guardrails, and version every change.
The payoff: more human conversations, lower latency, and far less time stitching infrastructure together. You build on the models you already trust. Pricing is transparent and flat at $0.08 per minute.
🔚 EXIT NODE
Three stories this week, one shared thread: AI infrastructure and control keep consolidating into fewer hands. Nvidia now touches the hardware most models train on and the open-source platform millions of developers build with. Anthropic showed its own agents can out-research human safety teams. Claude picked up another slice of your desktop. The tools got good enough to just start doing these things, and the permission structures are racing to catch up.
See you next issue.