「你无法提出你不知道的问题，你也无法创造你没有热情的产品。 这是一个复利循环：深度对话建构你所理解的 AI 的边界，而对话后的智慧成为你的创造杠杆。」写得太妙了。

Conversational AI is here. Build AI agents that can speak in minutes with low latency, full configurability, and seamless scalability.

I think the Deepseek moment is not really the Sputnik moment, but more like the Google moment. If anyone was around in ~2004, you'll know what I mean, but more on that later. I think everyone is over-rotated on this because Deepseek came out of China. Let me try to un-rotate

Last year, I joined DeepSeek with no RL experience. While conducting Mathshepherd and DeepSeekMath research, I independently derived this unified formula to understand various training methods. It felt like an "aha moment", though I later realized it was PG.

我对深度求索 #DeepSeek 的一份调查报告 在调查 DeepSeek 的过程中，我发现，想要真正了解其财务状况几乎是不可能的。DeepSeek

One more thing… We're integrating DeepSeek R1, a local model, into Craft's AI assistant as an experimental feature. Tomorrow, a select group of people will be able to download it to their devices and use it offline as well! Follow us for early access! 🚀 We believe AI should

We’ve raised a $180M Series C to give every AI agent a voice. The past year has been about building the foundations of AI audio - now we’re focused on making speech the standard for how we interact with technology.

WSJ把梁文峰的历史挖出来了。。 DeepSeek爆火这一周，硅谷还在持续破防中。漂亮国的人真是吓到了，WSJ又深挖了DeepSeek创始人梁文峰的一些细节，其中有一些之前的报道中没有提到，分享给大家。 全文翻译如下： 有人称他为中国的萨姆·奥特曼。 另一些人则把他比作量化投资的先驱吉姆·西蒙斯。

我们做了一个很有意思的东西。你可以用它来调整R1输出的速度，从而让自己的思考和r1的<think>同步，可以随时暂停修正你觉得R1思考错误的地方，把一些你的思想"植入"R1的思考过程中，最后得到最好的结果。似乎是R1特殊的训练方式，在其他reasoning模型上都很难做到这个。大家可以来玩玩。

假如优秀的端侧推理模型未来内置在各类 App 及手机系统里了，那么对于这类单一场景的 App 开发者来说，像 Dify、Coze 这类 LLM Ops 工具是否就没什么使用场景了？以及会有什么样新形态的工具或框架来进行端侧模型的运维呢？

RL 范式普及、推理模型平权后，应用爆发的周期即将开启，利好的还是 AAPL，看到一些投研在这个节点看空 iPhone 需求真是哭笑不得😂

🚨 o3-mini crushed DeepSeek R1 🚨 "write a Python program that shows a ball bouncing inside a spinning hexagon. The ball should be affected by gravity and friction, and it must bounce off the rotating walls realistically"

Trust Centered Design 信任感将变成产品设计时的重要思考原则

Poe 里 DeepSeek R1 和 V3 是同一个白菜价… 而且速度超级快根本不用等待，目前是由 together 和 fireworks 部署的😼

i'm so surfed with this Ghibli-style browser Deta i know this stunning branding style comes out way earlier than AGI(All Ghibli-style Image) GPT-4o image gen 🤣

Check out Kuse, a visual workspace that makes working and learning with AI intuitive. It analyses documents, searches the web, creates multimedia - and more! Try it for free: kuse.ai/?utm_medium=so…

Combining  Vision Pro + whisper + Warp + Figma MCP to literally “talk to design”, and watching the Wimbledon finals at the same time, is this the future of 10x designer?

After 2 years of using Cursor, 2023.7.14-2025.7.14, I decide to transfer to Warp and Claude Code. I’ve gradually shifted from Cursor to Warp, Claude Code, and Terminal-based workflows. Not because I’m a hardcore dev—but because something deeper has changed. With high-agency