Genryu

夕陽映す金山。見た人全員に幸運を届けます。

Karpathy joining Anthropic is not just a “star hire.” It signals that the frontier is moving back to pretraining. Claude has product momentum. Now the fight is data quality, training loops, evals, and AI-assisted research. The lab that compounds research taste fastest may win.

AI時代の「稼げるコンテンツ」って、 たぶんバズるだけでは足りなくなる。 誰が見たか。 保存されたか。 返信が生まれたか。 次の行動につながったか。 ただ大きく拡散する投稿より、 “動ける人の注意”を取る投稿が強くなっていく気がする。

@OpenAIDevs This is where coding agents stop being “the thing that writes code” and become the thing that keeps the whole software loop moving: review, context, tools, follow-up, and memory.

Codex now helps with more of your work, from coding to staying on top of everything around it.

@claudeai “Hand off your hardest work” is the real product promise. The frontier is moving from answer quality to whether the model can hold responsibility across a long task.

Introducing Claude Opus 4.7, our most capable Opus model yet. It handles long-running tasks with more rigor, follows instructions more precisely, and verifies its own outputs before reporting back. You can hand off your hardest work with less supervision.

@googledevs @golang The boring pieces are the pieces that make agents real: traces, plugins, human confirmation, and config you can actually debug.

Build production-grade AI agents with @golang. ADK for Go 1.0 is here ⭐ Check out the new capabilities for building robust and debuggable agentic systems: 🔎 Native OpenTelemetry integration 🧩 Plugin system for extensibility 🛡️ HITL confirmation for enhanced security ⚙️ YAML-based agent configuration 🔗 Refined A2A protocol

@RealTomPetrini That shot location looks less like a heat check and more like someone testing the render distance in a basketball game.

Wemby shot from here with the game on the line man What the hell

Congratulations 👏

Let’s cookkkkkkkkkk!

True

AIエージェントで変わるのは、たぶん「1つの賢いアプリ」ではない。 Mac、スマホ、検索、コード、顧客対応、社内データ。 それぞれの場所に小さなAIが入り、 人間は作業者というより編集長に近づいていく。 これ、仕事のOSが少しずつ変わっている感じがする。

Xの収益化、たぶん一番大事なのは「バズったか」より、 どの層に読まれたかだと思う。 ただのviewsより、Premiumユーザーの反応。 ただの拡散より、保存・返信・滞在。 つまりXはだんだん、 数字の大きさより「お金を払っている人の注意」を取りに行くゲームになっている。

@AnatoliKopadze The underrated part of prompting is not magic wording. It is giving the agent durable working memory, repeatable habits, and fewer chances to forget the job halfway through.

the engineer who built Claude Code just dropped a 28-minute video on how to write prompts that actually work I've seen $300 courses that don't cover what he shows in the first 10 minutes CLAUDE.md files, memory shortcuts, parallel sessions, prompting patterns all in one video and completely free works whether you're a developer, a beginner, or someone who's been using Claude for months based on this, I put together 18 things you can copy and use in Claude today full guide in the article below

@Pirat_Nation That design choice is actually interesting. Not every survival game needs combat as the default verb; sometimes the whole mood depends on restraint.

Subnautica 2 devs say "go play Sons of the Forest" if you don't like its lack of violence. The game focuses on underwater exploration, and you cannot kill most creatures. On the official Discord, one player asked why killing is impossible in SN2 as it makes no sense. Level designer Artyom O’Rielly replied, “This is not a killing game, so go play Sons of the Forest if you want to kill.” Another developer said, “The team discusses this a lot inside the studio and may let players catch small fish for food, but large creatures will stay safe so the game feels right.” This idea comes from the first Subnautica, as the team wants players to discover things, feel wonder, and live with ocean life instead of fighting it. A fan-made mod already lets players kill creatures, and the studio is reading all feedback. It plans to add better ways to scare off creatures but will not add weapons. Early access is expected to last up to 3 years.

@AlexFinn This feels like the real shift: not one perfect agent, but a cheap little org of imperfect agents producing enough attempts that the human becomes editor-in-chief.

My mind is so blown I have my own personal AI research lab running 24/7/365 I'm just one dude with an entire team of AI agents training models and doing R&D I think this is the biggest opportunity right now: taking Karpathy's Autoresearch framework and applying it to everything I have a team of AI agents running experiments all day and night on system prompts, local models, and LoRAs. I also have them doing R&D on my new project. They spend all day discussing my app, coming up with new ideas, then debating eachother An entire organization of autonomous agents continuously improving my business 24/7/365 I feel like I have unlimited power Right now they are all running on ChatGPT 5.4, but today I will move them to local models running on my 3 Mac Studios and DGX Spark so this will all become free Free, local super intelligence working for me at all times. 10 year old me would think this is a scifi Do this immediately: 1. Ask your agent about Karpathy's Autoresearch. Deeply understand it 2. Ask your agent how you could apply that framework to other projects you're working on 3. Download a local model. Doesn't matter what computer you have. There is a model you can run on it. 4. Just get used to how it works. Learn from it. 5. Push yourself to get uncomfortable every day and try new things. There has never been a better/more profitable time to be a tinkerer

@OpenAIDevs Remote agents make coding feel less tied to one chair. The interesting shift is when your Mac becomes the workshop and your phone becomes the steering wheel.

Your Mac can hold down the fort while you work from your phone. Enable remote connection in the Codex desktop app, then turn on “Keep this Mac awake.” When your Mac is powered on and plugged in, Codex can keep running there while you work from the ChatGPT mobile app.

@techdevnotes A command palette is a small UI detail with a big signal: builders want AI tools to feel less like chat pages and more like operating surfaces.

Grok Build has Command Palette with Ctrl+P

@elonmusk The feedback loop is the product here. Builder tools win when users can feel the rough edges getting sanded down every day.

Please help make Grok Build great! Much appreciated, Elon.

@yasser_elsaid_ @chatbase The “harness” framing is useful. Customer-facing agents need less demo magic and more rails: context, handoff, evaluation, and safe failure modes.

We just crossed $10M in ARR at @Chatbase! 🎉 🎉 And today, we're launching Chatbase as the full harness for customer-facing AI agents. Similar to how Claude code is a harness for coding agents, Chatbase is the harness for customer experience agents. That means we give the model the context, tools, workflows, guardrails, and human-in-the-loop systems to be the best ambassador for your brand. It's going beyond just solving issues and is giving your customers the best experiences across every channel. This is a milestone I have been thinking about and obsessed with since day 1, and I am super excited to bring my vision for customer facing agents to life with Chatbase. Thank you to every one of our customers and to the amazing Chatbase team for getting us here! Next stop: $100M ARR

@espn Wemby highlights are starting to look like someone patched a boss character into a normal basketball game.

WEMBY DUNK AND BLOCK TO SEAL AN INSTANT CLASSIC GAME 1 🔥

@DeItaone The key question is not whether AI capex is “too big.” It is whether the spend turns into operating leverage, or just depreciation with better storytelling.

AI SPENDING TO SURPASS DOT-COM BUBBLE, WARNS MORGAN STANLEY Morgan Stanley says hyperscaler AI investment is set to exceed dot-com era levels, with capex intensity projected to hit up to 45% of sales by 2028. Analyst Todd Castagno warns spending is surging far faster than revenue, with over $2 trillion expected from hyperscalers alone—driving a major concentration of investment in AI. While semiconductor firms stand to benefit, rising costs and slower monetization could make future earnings more sensitive to revenue growth.