#ComputerUse

Introducing agent-desktop agent-browser, for your whole computer → Computer Use for any coding agent → macOS, Windows and Linux → Runs in bg while you keep working → Agents use your computer in parallel, each with its own cursor

WARNING: The internet is addictive! ( my computer crashed 3 times making this one😭)

以防有人不知道！！ 现在任何 Agent 都能直接装上Computer Use！ Vercel Labs 正式开源了 agent-desktop，能给任何编程 Agent 加上桌面级的 Computer Use。 它和以前的做法不一样： 1、不靠截图！直接读系统的无障碍树，也就是界面上每个按钮、每个输入框的结构 2、在后台运行！不抢鼠标，你正常用电脑，它自己干它的！ 3、几个 Agent 可以各用一个光标，同时干活！ 4、支持macOS、Windows、Linux ！ 上手很简单，一行命令安装：npm i -g agent-desktop

Being a super intelligent computer is not always as great as it may seem 😂

@Risky_5l @grok 😪

"It's just a cute fox plushie on my computer desktop, right...?" (a Halloween teaser 👀) #SonkROM

Before → call an LLM, then build the tools and execution loop yourself. With Pine → integrate one SDK, send a job, get finished work back. Prompt in. Work out. #PineComputer #ComputerUse #DeveloperTools

AI is already smart enough. Real-world tasks are still slow, expensive and unreliable, because we hand AI a computer built for humans, then wrap it in a heavy harness. AI doesn’t need to get smarter. It needs a computer built for it. Today we’re releasing Pine Computer

🎉 Introducing OpenDuet Self-hostable, proactive AI agents that you own and works with ANY agent harness. Includes: - Build, generate, and edit live assets - Human-in-the-loop approvals - Computer use - Mobile, Web, and Channels Repo → Powered by @CopilotKit and AG-UI. Clone this template and customize it however you want. Enterprise-ready.

We’re opening a research preview of Claude Duet, a new app where Claude works as a proactive agent. It comes to you with what needs your attention, completes tasks you ask of it, keeps working when you step away, and checks with you before it acts on your behalf.

Everyone’s focused on making AI smarter. But what about the computer it uses? AI keeps advancing, while computers are still designed for humans. Pine Computer is built for AI, reading web pages as structured data rather than just screenshots. #PineComputer #ComputerUse #DeveloperTools

someone built a computer just for AI ✓ one sdk, send a job, get finished work back ✓ reads websites as structure, not screenshots this is what agents were missing FOR REAL #PineComputer #ComputerUse

If you want to use an LLM, you don't need a cloud subscription or massive server. You can run it right on your personal computer, and @manishmshiva shows you how in this guide. You'll learn how to install Ollama via the UI and command line, set everything up, and then run it locally.

*paws at you* can i infect your computer *paws at you* can i infect your computer *paws at you* can i infect your computer *paws at you* can i infect your computer *paws at you*

Wild how we made AI this smart and then gave it a computer built for humans. Pine Computer is a computer built for AI. A few lines of code and your agent has its own machine, browser open, ready to work. Watch it boot up👇 #PineComputer #ComputerUse

WE GAVE AI A PHD AND THEN HANDED IT A COMPUTER BUILT FOR HUMANS. this one is built for the model. 78.3% on SaaS-Bench, at ~1/20 the token cost of GPT-5.6 Sol + Codex. #PineComputer #ComputerUse #pineai

Every user gets their own cloud computer. One SDK handles execution. Your team picks the use case and ships the product. #PineComputer #ComputerUse

Watching Claude navigate desktop applications and analyze complex data in real-time is wild. Anthropic is building tools that don't just chat, but actually work alongside you. The era of autonomous AI agents is here. 🤖 #Anthropic #ClaudeCode #ComputerUse #FutureOfWork

This line sums up the whole idea: “We spent years making the model smarter. Now, we’re making the computer worthy of the model.” Pine Computer is a cloud computer built for AI, with the intelligence to use it built in. #PineComputer #ComputerUse #pineai

AI can write code in seconds. So why are companies still looking for great engineers? Because writing code was never the whole job. The real skill is knowing what to build, how to build it, and how to make it work in production. Here’s the roadmap to becoming an AI engineer in 2026 🧵 1. Master the Fundamentals Before jumping into AI, build a strong foundation. → Python, JavaScript, or C++ → Data Structures & Algorithms → OOP and problem-solving → Databases and APIs → Computer science fundamentals AI can generate code, but understanding it is what makes you a better engineer. 2. Learn Real-World Software Engineering Knowing how to code isn't enough. You need to know how to ship. → Frontend and backend development → Git and collaboration → Cloud computing → CI/CD and DevOps → System design and scalability → Security and performance This is where you stop writing code and start engineering systems. 3. Learn to Build With AI This is where the game changes. → AI coding assistants → LLM APIs and prompt engineering → Function calling and structured outputs → RAG and vector databases → AI agents and automation → Model evaluation and observability Don't just learn how to use AI tools. Learn how to build products with them. 4. Learn Production AI Engineering A demo that works once isn't a production-ready product. You need to understand: → Hallucinations and failure handling → AI security and guardrails → Testing and evaluation → Latency and cost optimization → Monitoring and reliability → Deploying AI systems at scale This is what separates AI experimenters from AI engineers. 5. Build Projects That Solve Real Problems Tutorials teach you concepts. Projects teach you engineering. Build an AI agent. Connect it to real tools. Add authentication, error handling, logging, and deployment. Then test whether it actually solves the problem. That's how you turn knowledge into experience. The biggest mistake you can make right now? Learning to prompt AI without learning how software works. AI is making coding faster, but engineering still requires judgment. The developers who thrive won't be the ones who write the most code. They'll be the ones who know what to build, understand what AI generates, and can turn it into reliable products. Learn the fundamentals. Master the tools. Build real things. That's the path to becoming an AI engineer. Save this roadmap for your learning journey.

Cursor Cloud Agents with Computer Use AI agents no longer just WRITE CODE — they RUN IT, TEST IT, CLICK around the UI, and SEND YOU a demo video with merge-ready PRs — no human in the loop 🔗 Full read: @cursor_ai #AIAgents #Computeruse #CloudAgents

「AIにPC操作させると、マウスを奪われて人間が作業できない問題」がついに解決しました。 新しく登場した『agent-desktop』が控えめに言って革命的です。 一言でいうと「agent-browserのPC全体版」。ブラウザの中だけでなく、PCのOS丸ごとAIエージェントの作業場に変わります。 これまでのComputer Use（AIによるPC操作）の弱点をすべて過去にする4つの特徴がこちら👇 ✅ 1. 作業を止めない「完全バックグラウンド動作」 従来のAI操作ツールは、画面もマウスカーソルもAIに乗っ取られるため、人間は画面を眺めて待つしかありませんでした。 agent-desktopなら、あなたが普段通りコーディングや資料作成をしている「裏側」で、AIが勝手に別ウィンドウやアプリを操作してくれます。 ✅ 2. 複数エージェントが「独自のカーソル」で同時作業 1台のPCの中に、複数のAI専用の仮想カーソルが存在するような状態です。 ・エージェントA：スマホシミュレータやElectronアプリを立ち上げて画面ポチポチ検証 ・エージェントB：Figmaのデスクトップアプリから素材を書き出し ・エージェントC：裏でブラウザとツールを行き来してエラー調査 これらが自分の作業を一切邪魔せず、同時並行で進みます。 ✅ 3. あらゆるコーディングエージェントに対応 特定のAIだけでなく、Claude Code、Cursor、Windsurfなど、普段使っているコーディングエージェントにそのまま「PC全体を動かす目と手」を拡張できます。 ✅ 4. macOS / Windows / Linux フル対応 特定の仮想環境だけでなく、普段使っているメインOS上でそのまま動くのがデカすぎます。 💡 具体的に何が起きるか？ これまでは「ターミナルで完結する作業」しか完全自動化できませんでしたが、これからは「コード修正 → アプリ起動 → AIが自分で画面をクリックしてE2Eテスト → UI崩れやバグを検知して再修正」というループが、自分の作業の裏で勝手に完結するようになります。 「人間がAIの操作を待つ」時代から、「人間が仕事している裏で、複数のAIが別々のアプリを同時に動かして働く」時代への転換点になりそうです。 ところで、「どうやって人間のマウスを奪わずに裏で別々のカーソルを動かしているのか？」という技術的なタネ明かしと、実際の開発環境への組み込み方気になりませんか？ 参照動画： 続きのディープな仕組みは @grok にわかりやすく解説してもらいます👇 @grok 『agent-desktop』のように、OSの物理マウスカーソルを奪わずにバックグラウンドで複数のAIエージェントが独立してGUI操作できる技術的な仕組み（OSごとの裏側の挙動）と、Claude CodeやCursorと組み合わせた時に開発者が一番時間を削減できる具体的な活用シナリオを、初心者にもわかりやすく噛み砕いて教えて！