Agent PM/Dev、INTJ｜Obsidian、Codex、Figma｜Context、AGUI、UXUI Design https://teapawt.net/t.co/mItUvHD69g
北京
Joined January 2020
- Tweets2.8K
- Following1.2K
- Followers877
- Likes11.4K
Sam Song retweeted
Tim 送 100 台 iPhone 涨粉 27 万，我心态崩了。
徐跑跑抽 10 台 iPhone Duo，我彻底崩了。
那薄肌教父也发个福利🎁 本次抽 15 台 iPhone Duo 2TB
人人都有机会，不看粉丝多少，不需要有蓝标。
参与方式：
关注我，一键三连本帖
并在评论区回复💬
24h 以后我让 @grok 抽奖🎟️ 保证公平公正公开。
不要等别人抽到了，才后悔没参加。
抽 15 个薄肌小子送给我，人人都有机会🙂🙂
Sam Song retweeted
gaocegege.com/Blog/genai/ser…
最近太累，精神也不好，没有什么精力写文章
Sam Song retweeted
I’ve always found that the way to get to an elegant design is to go way overboard and edit back.
Like shooting 50x the footage you need and cutting a movie from it.
The best designers and engineers know this. They expect to throw away *a lot* of ideas and implementations.
The most frustrating people to work with are those who are precious about every single keystroke.
Now the people who love to FAFO are supercharged by coding agents. They’ve built the muscle memory and now they have a mech suit to go absolutely nuts.
Sam Song retweeted
I found a site that solves pretty much every UI component oversight problem.
It gives you exact, step-by-step checklists for building cards, navigation, and forms so you never miss critical micro-details again.
Link: checklist.design
Sam Song retweeted
Given some of the results I'm seeing recently, it's pretty clear Codex is a good harness.
But it will seem primitive in 2-3 months and we're about to go through another major evolution in how we use AI at the frontier. The next generation of models need more than your laptop.
Sam Song retweeted
GPT 5.6 Sol Max 写代码非常的防御，属于脑子不是那么灵光，但是劲大。你裤子漏个洞，他通常选择掏出一块钛合金帮你补上，你告诉他应该用块布来补，他会说“对，这样更好”，然后飞快把钛合金拔掉，用布补好。
Sam Song retweeted
Most people don't hate working.
They hate working on things that don't matter, for people who don't notice, toward goals they don't care about.
Sam Song retweeted
i think there’s 2 categories of graphs to think about here
1. static graphs that manage agent control flow, like the core model and tool calling loop embedded in a larger goal loop
2. dynamic graphs that the agent generates to solve a task, like code the agent can write to fanout and synthesize results at runtime
both are important!
if this resonates w folks, happy to share more thoughts
Sam Song retweeted
In 40 minutes, @sydneyrunkle and @jakebroekhuizen run through everything you need to know about Deep Agents.
What are Deep Agents?
How do they work?
What are some of the most common use cases?
What are the core components?
How do file systems work?
Watch the full session ⤵️
Sam Song retweeted
@mulerun_ai 一个UGC 的 Agent 超市，类似 coze 的商店。
做了一个产品设计：把每个 Agent 调用的消耗点数外显，真正解决了 Agent 2C商业化的一个探索。
相当于为 Agent 创作者构建一个发布平台。好的 Agent 是可被商业化的。
mulerun.com/
Sam Song retweeted
Ethereum’s largest ecosystem fund. LINEA’s token echoes the distribution of the Ethereum Genesis claim and puts the ecosystem first, directing the majority of funds toward users, builders, Ethereum public goods, and R&D.
85% Ecosystem (75% development fund + 10% early users)
15% Consensys treasury (5-year lockup)
No insiders. No value extraction. Just Ethereum.
Sam Song retweeted
Claude now has a 1M+ context window!
@AnthropicAI absolutely cooked with the subagent release!
@rasmic was right when saying to bet on the model provider building their own tooling
Combined everything I've learned from
@elie2222 @iannuttall @EyalToledano to build a beast!
Readers added context they thought people might want to know
The author admits that Claude doesn’t actually have a 1M token context window.
x.com/AbdoMobayad/st…
Sam Song retweeted
andrej karpathy 谈“如何成为任何领域的专家”：
（又一个“纯金”建议💛）
1 迭代式项目驱动：追求深度而非广度，“按需”学习（不是先学后做学完再做，而是做中学，根据项目需要边做边学）
2 费曼一切所学所做：用自己的话说出来/总结
3 永远不要（和他人）横向比较，只（和过去的自己）纵向比较。
---
1 的本质是项目式学习
2 的本质是费曼技巧
3 的本质是成长思维
这些都是大家听说过的道理。差异在于大多数人道理都知道但是做不到🙁
