夜雨聆风📋 摘要速览
🎓 小白解读
今天的 AI 圈像一场"能力大跃迁"的接力赛。Claude Code 的 Workflows 让 AI 编程助手不再只会写代码，还能编排复杂工作流——这意味着非程序员也能指挥 AI 干活了。OpenAI 的 Codex 数据更直观：知识工作者的采纳速度是开发者的 3 倍，数据分析类任务周增长 110%——AI 正从极客工具变成全员工具。
Vercel CEO 的 "YES-CODE"宣言值得细品：过去"低代码"兴起的前提是"代码贵且难"，现在 Agent 让代码又便宜又多，整个品类逻辑被翻转。而 Peter Yang 的观察则指向另一面——如果你的 SaaS 只做一件窄事，AI Agent 凭上下文和记忆就能干得更好，$20/月的订阅根本扛不住和 ChatGPT 订阅的价值对比。
Google 的 Thinking Levels 和 OpenClaw 进企业则是基础设施层的进展：前者给用户控制模型"想多深"的权力，后者把个人 AI 助手带入合规的企业环境。Sam Altman 对美国新行政令的表态则表明，行业大佬正积极拥抱"发展+安全"双轨制。
📖 原文精选
🗣️ Claude Code Workflows：Skills 之后最大升级
Workflows are the biggest upgrade to Claude Code's capabilities since skills and subagents. I dove deep into it with @sidbid to figure out best practices, examples and more. I'm particularly excited about the non-technical tasks it enables for Claude Code.
Claude Code 的 Workflows 是自 Skills 和 Subagents 以来最大的能力升级。Thariq 与 Sid Bid 深入探讨了最佳实践和用例，特别令人兴奋的是它为 Claude Code 开启了非技术任务的可能性。
📎 Thariq · X · 2026-06-02 · 查看原文[1]
🗣️ Codex 知识工作者增速碾压开发者
From OpenAI's latest Codex report: "Knowledge workers now represent about 20% of Codex users and are adopting it more than 3 times as fast as developers. The fastest growing task types for knowledge workers are: Data Analysis (110% growth week over week); Research (+37%), and Knowledge Artifacts (+36%)"
OpenAI 最新 Codex 报告：知识工作者已占 Codex 用户的 20%，采纳速度是开发者的 3 倍以上。增长最快的任务类型：数据分析（周增 110%）、研究（+37%）、知识产出（+36%）。
📎 Zara Zhang · X · 2026-06-02 · 查看原文[2]
🗣️ YES-CODE：代码不再稀缺，低代码的底层假设被颠覆
YES-CODE. An entire category of software, "no-code", was built under the presumption that code is expensive, difficult, and scarce. Coding agents have forever changed the equation. Code is now cheap, easy, and abundant. Our mission is to create the easiest cloud for agents that you never graduate from.
YES-CODE。"低代码"这个品类建立在"代码昂贵、困难、稀缺"的假设之上。编程 Agent 永远改变了这个等式。代码现在变得廉价、简单且丰富。我们的使命是为 Agent 打造你永远不需要"毕业"的最简云平台。
📎 Guillermo Rauch · X · 2026-06-02 · 查看原文[3]
🗣️ Gemini 全平台上线 Thinking Levels
✅ Papercut fixed: Thinking Levels are now available on Gemini across Web, iOS, and Android.
已修复！Thinking Levels（思考深度调节）现在在 Gemini Web、iOS 和 Android 端全部可用。
📎 Josh Woodward · X · 2026-06-03 · 查看原文[4]
🗣️ OpenClaw 进军企业级市场
Such a privilege to work with Microsoft to bring claws to enterprises!
非常荣幸与微软合作，将 Claws 带入企业！
📎 Peter Steinberger · X · 2026-06-02 · 查看原文[5]
🗣️ 窄场景 SaaS 正被 AI 替代
People are saying SaaS is not dead. I think larger enterprise SaaS that can do multiple jobs are probably fine. But if you're building a simple SaaS for a narrow use case, I think it's harder to monetize now because AI skills can often solve the same problem in a much more flexible, personalized way. People are willing to pay hundreds or thousands for services but charge $20/month for a SaaS and people will compare its value to their Claude/ChatGPT subscription.
人们说 SaaS 没死。能做多种事的企业级 SaaS 可能还行。但如果你在做窄场景的简单 SaaS，现在更难变现了——AI Skills 能以更灵活、更个性化的方式解决同样的问题。人们愿意为服务付几百几千，但你的 SaaS 收 $20/月，人们就会拿它和 Claude/ChatGPT 订阅比价值。
📎 Peter Yang · X · 2026-06-02 · 查看原文[6]
🗣️ Sam Altman 谈美国 AI 行政令
The US should lead on AI by continuing to develop the very best models, making sure they're safe, and getting cyber tools into the hands of trusted defenders. The new EO gets the balance right.
美国应在 AI 领域保持领先——持续开发最优秀的模型、确保安全、将网络工具交到可信的防御者手中。新的行政令在平衡上做得对。
📎 Sam Altman · X · 2026-06-03 · 查看原文[7]
🗣️ 语言就是新时代的 API
This is what education in the age of AI looks like. Start with the language: the linguistic surface is your roadmap. English alone could not produce tangible things. You had to translate it into machine instruction. You can now go direct.
这就是 AI 时代的教育。从语言开始：语言表面就是你的路线图。过去仅靠英语无法产出实体成果，你必须翻译成机器指令。现在你可以直达。
📎 Guillermo Rauch · X · 2026-06-02 · 查看原文[8]
🗣️ Token 成本催生模型路由层
As token budgets take on a larger part of operating expenses over time, model routing is the inevitable conclusion. The products that can intelligently route these workflows to the right tier of model will be in a strong position to aggregate more demand.
随着 token 预算在运营支出中占比越来越大，模型路由是必然结论。能智能地将工作流路由到合适模型层级的产品，将在聚合需求方面占据强势地位。
📎 Aaron Levie · X · 2023-06-03 · 查看原文[9]
🗣️ SWE 基准测不出应用构建能力
SWE benchmarks don't necessarily capture app building capabilities. ViBench does.
SWE 基准测试不一定能捕捉应用构建能力。ViBench 可以。
📎 Amjad Masad · X · 2026-06-02 · 查看原文[10]
🗣️ 推理效率的最佳奖励函数
Probably the best reward function for reasoning efficiency I've seen.
这可能是我见过的最好的推理效率奖励函数。
📎 Swyx · X · 2026-06-03 · 查看原文[11]
🗣️ 法律 AI 公司 Legora：每次新模型都在涨潮
Interpreting law is one of the oldest jobs in the world. @MaxJuneststrand, co-founder and CEO of @WeAreLegora, is bringing it into its next era with Claude. His bet: every new model release raises the tide, and Legora is building the boats for everyone else.
法律诠释是世界上最古老的职业之一。Legora 的联合创始人兼 CEO Max Junestrand 正在用 Claude 将其带入新时代。他的赌注：每次新模型发布都在涨潮，而 Legora 在为所有人造船。
📎 Claude · X · 2026-06-02 · 查看原文[12]
🗣️ Opus 4.8 评价两极化？
Almost a week later! What are your thoughts on Opus 4.8? We were extremely bullish on it in testing—it seems the response was more tepid once y'all got your hands on it. One theory I have is that by nature it pushes on your frame a little more, and the results are high-variance—sometimes it does something amazing, and sometimes it disagrees in a way that is obviously wrong.
快一周了！大家怎么看 Opus 4.8？我们在测试时非常看好——但你们上手后似乎反应更冷淡。一个理论是它天生更"反骨"，结果方差很大——有时做出惊人的事，有时以一种明显错误的方式反驳你。
📎 Dan Shipper · X · 2026-06-02 · 查看原文[13]
🗣️ 听见用户：Listen Labs 用 AI 同时跑数千场语音访谈
Our goal is to get to a billion people in our audience and then to be able to stratify and know what exactly is this person an expert on. As we get closer to AGI, it will be easier to build things, but the hard part will be knowing what to build.
我们的目标是让受众达到十亿人，然后能分层精确知道每个人擅长什么。越接近 AGI，构建越容易，但难的是知道该构建什么。
📎 Alfred Wahlforss · Podcast · 2026-06-02 · 查看原文[14]
🗣️ 创始人融资门槛从未更高
The best founders treat the following as a necessary component of their business but NOT the entire business: AI/timing/why now, Funding, Distribution/hype, Market, Product, Revenue. The bar has frankly never been higher and you really need to be able to explain how you're capturing at least multiple of these in a unique way.
最优秀的创始人把以下每一项视为必要组件，但不是全部：AI/时机、融资、分发/声量、市场、产品、收入。门槛从未如此之高，你必须解释如何以独特方式同时拿下多项。
📎 Nikunj Kothari · X · 2026-06-03 · 查看原文[15]
📖 今日术语
| English | 中文 | 解释 |
|---|---|---|
| Workflows | 工作流 | Claude Code 新功能，允许编排多步骤 AI 任务流程，超越单次对话模式 |
| Thinking Levels | 思考深度 | Gemini 新功能，让用户控制模型推理的深度层级，平衡速度与质量 |
| Model Routing | 模型路由 | 根据 token 成本和任务复杂度，智能将请求分配给不同层级的模型 |
| ViBench | Vi基准 | Replit 提出的应用构建能力基准，弥补 SWE 基准无法衡量完整应用构建的缺陷 |
| YES-CODE | 是代码 | Vercel CEO 提出的理念，代码因 Agent 变得廉价丰富，低代码的前提不再成立 |
参考链接
[1] 查看原文: https://x.com/trq212/status/2061907538741006796
[2] 查看原文: https://x.com/zarazhangrui/status/2061924300698091760
[3] 查看原文: https://x.com/rauchg/status/2061934154732974376
[4] 查看原文: https://x.com/joshwoodward/status/2062025667852812583
[5] 查看原文: https://x.com/steipete/status/2061874084649025728
[6] 查看原文: https://x.com/petergyang/status/2061846283263103274
[7] 查看原文: https://x.com/sama/status/2061973280655904815
[8] 查看原文: https://x.com/rauchg/status/2061862134469062850
[9] 查看原文: https://x.com/levie/status/2061974298760495132
[10] 查看原文: https://x.com/amasad/status/2061878314311266552
[11] 查看原文: https://x.com/swyx/status/2062060142489973010
[12] 查看原文: https://x.com/claudeai/status/2061829558999912680
[13] 查看原文: https://x.com/danshipper/status/2061817375519809665
[14] 查看原文: https://www.youtube.com/watch?v=Rumft-rsEu4
[15] 查看原文: https://x.com/nikunj/status/2062033620773306763
