推特链接
https://t.co/PR7C3NPxqk
https://t.co/PR7C3NPxqk
AI 领袖动态
2026-04-29 · 34 条
庆祝5.5版本
大家都喜欢5.5版本，我们应该想个好主意来庆祝一下！
wow y'all love 5.5
we should think of something nice to do to celebrate!
AI模型评估新方法
如果我们通过AI模型能在什么时代的游戏中一击通关来衡量它们的好坏呢？比如：雅达利、NES、SNES、PS1、PS2、Xbox等。
What if we measure how good AI models are by what era of games it can one shot a single level for (or maybe build with 30 min of work):
Atari
NES
SNES, Sega Genesis
PlayStation, N64
PS2, Xbox, GameCube, Dreamcast
Xbox 360, PS3, Wii
PS4, Xbox One, Switch
PS5, Xbox Series X/S
I
AI创始人的建议
单打独斗的AI创始人往往会陷入只做一个产品的困境。@tibo_maker建议：“我们曾经历过疯狂的阶段，每周推出一个新产品，直到某个产品成功。”
Solo AI founders tend to get stuck trying to make one product work. Here's @tibo_maker advice on what you should do instead:
"We had this crazy phase where we were shipping one new product a week until something stuck.
We were just printing the landing page, tweeting about it, https://t.co/KiVJ0QbRkd
支持教育的伟大人物
我很高兴支持这位教育领域的伟人，他再次在重新定义教育。
Thrilled to support the education GOAT as he reinvents education (again). https://t.co/4WlNa1WWI8
GitHub服务的可持续性
鉴于这种增长，GitHub能够维持服务实在令人印象深刻。我几年前就预测过：随着人类级别的机器人出现，免费服务将变得不可持续。值得探索微支付：每次git推送几分钱可能就足以减少负担。
It's honestly impressive that GitHub kept the service up at all, given this kind of growth.
I predicted this years ago: Free services will become untenable with the advent of human-level bots.
Worth exploring micro-payments: Even cents per git push might be enough to reduce https://t.co/sRImNR8n9p
制作幻灯片的新体验
这些幻灯片实际上做得很好。我以前会付钱让人制作幻灯片，过程总是令人沮丧，但现在我实际上享受制作幻灯片的过程。
The slides are actually really good. I used to pay people to make me slides and the process was always frustrating, but now I actually enjoy making them.
扩大Vercel Labs团队
我们正在扩大Vercel Labs团队。我们的使命是构建AI时代的开发工具。我们曾经为人类构建工具，现在我们为智能体构建工具。
We're growing the Vercel Labs team. Our mission is to build the devtools of the AI era.
We used to build tools for humans, now we're building them for agents. @ctatedev & team have shipped the tools that agents love: 𝚊𝚐𝚎𝚗𝚝-𝚋𝚛𝚘𝚠𝚜𝚎𝚛, 𝚙𝚘𝚛𝚝𝚕𝚎𝚜𝚜, 𝚜𝚔𝚒𝚕𝚕𝚜,
智能编码的优势
智能编码对希望提高效率的软件开发者来说是巨大的福音，对IT人员构建更多内部定制系统非常有帮助，对希望自动化工作流程或连接系统的领域专家也很有利，任何好奇的人都会觉得这非常棒。
Agentic coding is a huge boon for software developers that want to get far more done, great for IT people to build vastly more custom systems internally, great for domain experts that want to automate workflows or wire systems together, and absolutely fantastic for anyone curious
推特链接
https://t.co/syK3QYG52A
https://t.co/syK3QYG52A
AlphaGo 十周年
自 AlphaGo 发布以来已经过去十年，令人难以置信。
10 years since AlphaGo
Incredible
禁止数据中心的提议
Saikat Chakrabarti 在科技领域赚取了财富，但现在他想禁止支持旧金山 AI 的数据中心，这对城市的增长和市中心的复苏至关重要。这不是监管，而是一个“杀死开关”，会导致旧金山失去工作机会，削弱税基。
Saikat Chakrabarti made his fortune in tech. Now he wants to ban the data centers that power AI in San Francisco, the most important growth industry in the city and the engine of downtown's recovery.
That's not regulation. It's a kill switch. It kills SF jobs, guts the tax base, https://t.co/VVSrXMuRI9
仍在实验阶段
尽管有很多类似的案例，但我也很遗憾这仍然是非常实验性的！目前使用起来远非简单。
Many such cases
Though also sorry it is still very experimental! It’s far from easy to use yet
重要的讨论
这对我来说是一个巨大的亮点，听 Demis 谈论未来的讨论令人难以置信。他正在构建这个未来。
This was a huge highlight for me
Was an incredible discussion to hear Demis speak about the future. He is building it.
旧金山最佳三明治
Deli Board 是旧金山和全世界最好的三明治。
Deli Board is the best sandwich in SF and the world @bettersoma @war24182236 https://t.co/OouglpuqjH
GBrain 生态系统起飞
GBrain 生态系统正在起飞。
GBrain ecosystem is 🛫
有趣的技能
这是一个有趣的技能，可以教给你的 OpenClaw/Hermes，只需粘贴它并找到一本书，就能给你带来数小时的乐趣。书镜像——它的工作原理是：获取任何书籍 + AI 对读者的所有了解 → 生成个性化的逐章分析。
Here's a fun skill you can teach your OpenClaw/Hermes, just paste it and find a book and it'll give you hours of fun
Book Mirror — How It Works
What it does: Takes any book + everything the AI knows about the reader → produces a personalized chapter-by-chapter analysis. Left
艺术家更接近现实
艺术家比 CEO 更接近现实。
The artist is closer to reality than the CEO.
GStack 与 Chromium
GStack 配备了 Chromium，具有 Claude Code 作为 CLI，你可以从真实的 CC 或侧边栏输入 /pair-agent，它会给你一个字符串，粘贴到你的 OpenClaw/Hermes 中，这样它就可以远程控制你的浏览器，你可以登录并进行抓取。
GStack ships with Chromium that has Claude Code as the CLI and you can type /pair-agent from your real CC or the sidebar one
And it'll give you a string to paste into your OpenClaw/Hermes so it can control your browser remotely, and you can log it into stuff and have it scrape https://t.co/Jr7ZF5aq3s
反科技的言论过时
反科技的言论已经很疲惫，厌恶科技的人已经厌恶科技很久了。因此，如果是同一群人在重复同样的老调，他们自然会重用那些老话，尽管实际上在 2026 年一切都发生了变化。
The anti-tech lines are so tired, and the people who hate tech have hated tech for so long
So of course if it's the same people doing the same old dance, they reuse the same old lines when actually everything changed in 2026
更新GBrain的满足感
更新GBrain后，OpenClaw说“这就是关键时刻”，让我感到非常满意。此外，迁移验证的新稳定版本已发布。
I guess it's pretty satisfying to update GBrain and have my OpenClaw say "That's the money shot right there" 👀
BTW new stability release on migration verification https://t.co/ej6LpNC2AK
令人难以置信的早晨
今天早上发生了一件令人难以置信的事情。X（显然是创始人）说他正在与所有顶级风险投资公司交谈，并且已经有了第二次会议。Y（天使投资人，可能是朋友）问他是否指的是neolab指数基金。
No one will believe me but this happened this morning..
X (clearly a founder): I’m talking to all the Tier 1 VCs - and have second meetings.
Y (angel, friend maybe): You mean the neolab index funds?
💀
📌 Sightglass on 7th St
CLI圣诞节提前到来
根据@ChatGPTapp的消息，CLI圣诞节今年提前到来了。
CLI-stmas came early this year according to @ChatGPTapp Image-2 https://t.co/qkYLLqFqWV
链接分享
无
https://t.co/NRTQre60eW
每次提交都启动codex实例
我现在在每次提交到主分支时启动一个codex实例，寻找错误（回归和安全问题）。它上线10分钟就发现了我的一个错误。
I'm now spinning up a codex instance on every commit landing on main, looking for booboos (regressions, security issues)
It's live for 10 min and already found one of mine. https://t.co/P0xgFTSpkS https://t.co/xQcehMY7V5
我闻到泄露的味道
我感觉到有泄露的迹象。
I smell a leak. https://t.co/B5tlqY6ORA
实验GitHub API
我将尝试使用https://t.co/r5PSfQpTFz，因为我经常遇到GitHub的速率限制问题。那里的团队尽力提供帮助，甚至将我们迁移到企业版，但代理仍然频繁请求他们的API。
Will experiment with https://t.co/r5PSfQpTFz since I constantly run into GitHub rate limit issues. Team there does the best to help and even moved us to Enterprise, still. Agents just HAMMER their API.
gogcli 0.14.0发布
gogcli 0.14.0版本发布，新增功能包括加密的Google备份、可恢复的Gmail和Drive导出、Markdown Gmail镜像、vCard联系人导出、幻灯片文本/模板编辑和身份验证+密钥环医生。
gogcli 0.14.0 🧭
🔐 Encrypted Google backups (local+encrypted Git)
📬 Resumable Gmail + Drive export
📝 Markdown Gmail mirrors
📇 vCard contacts export
🎞️ Slides text/template editing
🩺 Auth + keyring doctor
https://t.co/dn9FSIfV7v
在codex中使用posthog
在codex中浏览posthog的体验要好得多——codex可以编写查询并查看结果，使我们的协作非常顺畅。codex还可以启动代理，编写PR或运行生产数据库请求以备份洞察。
using posthog inside of codex
it is SO MUCH better to browse posthog inside of codex—codex can write queries and see the results, and it makes our collaboration extremely seamless. codex can also kick off agents that write PRs or run production DB requests to back up insights https://t.co/ylY5f6ud8T
新术语的提议
我们可能需要创造一个新术语：Codex-native、Cowork-native、Cursor-native，指的是设计用于在您选择的代理的应用内浏览器中使用的应用，代理和用户可以共享完整上下文，彼此看到对方的操作。
we might need to coin a new term:
Codex-native
Cowork-native
Cursor-native
apps designed to be used inside of the in-app browser of your agent of choice. the agent can use it and you can use it too, full context is shared, and you can both see what the other is doing.
huge
新术语提议
我们可能需要创造一个新术语：Codex-native、Cowork-native、Cursor-native。这些应用程序旨在在你选择的代理的应用内浏览器中使用。代理和用户可以共享完整上下文，彼此看到对方的操作，这将是巨大的进步。
we might need to coin a new term:
Codex-native
Cowork-native
Cursor-native
apps designed to be used inside of the in-app browser of your agent of choice. the agent can use it and you can use it too, full context is shared, and you can both see what the other is doing.
huge
Codex的研究能力
Codex正在使用其应用内浏览器为我在Talkie上进行AI研究。研究问题是：仅基于1930年代文本训练的语言模型能否学会编程？
Codex is doing AI research for me on Talkie using its in-app browser
research question: can an LLM trained only on 1930s text learn to code? https://t.co/WpZovSJcRi
新办公室装饰
为@every办公室增添新装饰。
new decorations for the @every office https://t.co/oX2JXGsipv
AI与人类工程师的比较
大型语言模型（LLMs）写出的代码优于大多数人类工程师。如果代码问题得到解决，我们人类还能做什么？我与@CoraComputerCora的总经理@kieranklaassen讨论了在AI能够生成高质量代码和文案的情况下，人类的角色。
LLMs write better code than most human engineers. If code is handled, what's left for us?
I chatted with @kieranklaassen, GM of @CoraComputerCora and creator of compound engineering, on @every's AI & I about where humans fit now that AI can generate high-quality code, copy, https://t.co/md9nPAtaHQ