If agents end up writing most of the code, do we still need this many languages and frameworks? They don't care what they write in. One well-designed language, built by the best minds and improved over time, would be far easier to upgrade and keep secure.
I've been cutting apps on my iPhone as Grok Bot takes over what they used to do. Credit Karma can go now too.
One more down.
xitter.cf/elonmusk/status/211296…
Been uninstalling apps on my iPhone, hoping to end up with just the few I really need. Grok Bot and other AI agents keep getting better at the things I used to need an app for.
For all the talk about Mars, Earth still has views that stop you mid-sentence. Some places just remind you how lucky we are to be here.
iPhone screens are almost unreadable in direct sunlight. Walk from shade into the sun and even max brightness barely helps. Meanwhile a Kindle looks better the brighter it gets. Feels like phones could learn something from e-ink.
Been reading about informal networks in n organizations. The people everyone goes to for advice are often not who the org chart would tell you. Makes me wonder how that changes when half the questions go to an agent first.
That question just got a lot less hypothetical. xitter.cf/ytk_matsuda/status/211…
イーロン・マスク「人間の脳・記憶・人格をテスラロボット（Optimus）にダウンロードすることは可能だ」
イーロン・マスクとレイ・カーツワイル
「肉体が滅びても、AIロボットに自分を移せば『永遠の生命』になる」――世界一の富豪が本気で語る「デジタル不老不死」と人類サイボーグ化の衝撃
「『死んだらすべて終わり』という人類の絶対法則を、テスラとロボット工学が破壊する」
イーロン・マスクが独メディアのインタビューで明かした、「ロボットへの意識ダウンロード」と「すでに始まっているシンギュラリティ」の全貌をわかりやすく👇
・「自分の記憶・個性」をOptimusに丸ごと移植できる
「人間の脳の能力を人型ロボット『Optimus』にダウンロードできるか？」という問いに、マスクは「可能だと思う」と即答。肉体を離れることによる感覚の差はあるとしつつも、「自分を自分たらしめている独自の記憶や個性を保存・移送することは可能だ」と明言。
・生身の肉体を捨てる「新しいかたちの不老不死（永遠の生命）」
心臓が止まり肉体が朽ち果てても、自分のデータと人格がテスラのロボットの中で生き続ける。宗教やオカルトではなく、工学的なアプローチとして「死の克服（デジタル不老不死）」が現実の射程に入った。
・「人類はすでにサイボーグ化している」という事実
カーツワイルが予言した「シンギュラリティ」について、マスクは「ある日突然ハッキリ訪れるものではない」と指摘。すでに人類はスマホやPCに写真や動画（外部記憶）を預け、脳の能力を異常なレベルで拡張している。地球の裏側と無料でリアルタイム通話し合うなど「300年前なら魔女として火炙りにされていた魔法」を誰もが日常で使いこなしている。
・ 「全人類80億人の脳みそを全部足しても、コンピューターに秒で負ける」
地球上の人間全員の頭脳をフル回転させたパワーより、世界中のコンピューターが叩き出すパワーのほうが圧倒的に巨大になってしまった。生身の肉体に入った人間の脳なんて、もはやデジタルの知能の足元にも及ばない。
要するに、「『いつかシンギュラリティが来る』のではなく、人間はすでにスマホで脳を外部拡張したサイボーグになっており、次の段階として【自分の記憶や人格をテスラのヒト型ロボットに丸ごとダウンロードして永遠に生き続ける】というSFの世界を、イーロン・マスクはガチの工学的現実として予測はしている」ということです。
Small thing I love about Grok Bot. I tap a thumbs up or down and it actually gets it, and it even reacts back with its own emoji. Feels less like using a tool and more like texting a teammate. Tiny detail, big difference in how working with agents feels.
Bezos loved The Expanse so much he saved it from cancellation. I'd love to see someone do that for Asimov. Foundation and the Robot series, told faithfully and connected the way he wrote them. Apple's Foundation was fun, but it drifted a long way from the books.
It is good to remember this. Coding agents are eating the entry-level tasks. That doesn't mean we stop hiring juniors. It means we teach them to review, debug, and own outcomes from day one. Tomorrow's senior engineers are being trained right now, or they aren't.
Things are becoming a compute race.
Pairing the biggest ambition with the best chips = faster future. xitter.cf/farzyness/status/21118…
Agree. The scarce skill is knowing what's worth building. xitter.cf/dhh/status/21116370458…
agents that survive restarts. Meanwhile i still can't resume a dream at 3am.
Harness durability in fx.sh.
The same 𝚌𝚛𝚎𝚊𝚝𝚎𝙵𝚡𝙰𝚐𝚎𝚗𝚝() is now *durable* on your laptop or in the cloud, via files, Postgres, or Vercel Workflows. Node.js or Bun.
Agents need long-running compute. But they also need to tolerate restarts, redeploys, timeouts, crashes…
Concretely, if a user sends a prompt and then throws their phone into a black hole, we ensure that turn runs to completion. If the process crashes mid-tool-call, it resumes from there.
To this end, 𝚕𝚒𝚋𝚏𝚡 introduces a simple durability adapter API. On Vercel, it leverages all the infra we built for durable workflows and queues, giving you reliability with 1LOC.
This is a critical aspect of production-grade agents (and something all @eve agents also get out of the box). Next, we'll show you how we bring the same durability to Pi, OpenCode, and even your own custom harnesses.
Design first, then let it build. Same rule i'd give a new engineer.
How to do you use AI coding agents for production codebases (not vibe-coding) ?
Here are the 4 setups I tried for real client work, why 3 failed, and the one that stuck: align with the agent on the design first, then let it implement.
00:00 Introduction and Disclaimer
00:53 Vibe Coding vs Production Software Engineering
03:20 One-Shotting Features with Claude Code
04:21 Planning Skills and Vertical Slicing
05:55 Herder for Running Many Agents
07:14 HumanLayer IDE and Task Setup
09:21 Research, Design, Implement Workflow
Enjoy
60% of deploys made by agents. The job isn't writing the code anymore, it's deciding what's allowed to ship.
Some interesting machine & agent Vercel traffic stats.
• Across the entire Vercel network, 58.18% traffic is bot-originated (last 30d). This was 32% in Jan 2024.
• 60%+ of deployments on Vercel are now agentic, up from ~3% in Jan 2026.
• Up to 83% of pageviews are now coming from agents to Vercel's own documentation sites. The % reliably increases the more we optimize content for them.
I expect 'direct' human internet traffic to basically become a rounding error in coming years. The web will thrive, but it will be built for and by agents.