推文
357 tweets · 110 sources账号:全部mattpocockukdair_aisimonwbchernyadocompletetrq212felixriesebergdani_avila7dexhorthyyetonezarazhangruibadlogicgamespetergyang0xblacklightctatedeviamzhihuikarpathykunchenguidmckaywrigleyryancarsontheoaidenybaidoteygarrytankieranklaassenleon7haoomarsar0thdxr9hillsGeminiAppRLanceMartinRhysSullivanZaynHaodillon_mulroylennysanmvanhornnummanalivikingmuteyiliush0xPauliusAYi_AInotesAlchainHustAnthropicAIBenJames_____ChadMoranClaudeDevsDanielMiesslerDaveJDavidKPianoGoogleDeepMindHilaShmuelJack_W_LindseyJiaxi_CuiLinghuaJMeari_V2_0_GMnilaxRaillyHugoTaniyatweets_addyosmaniaibuilderclub_alliekmillerandrewfarahantirezartmanasynkimobbssppllvvcailynyongyongcgtwtscharmaine_kleechrisbarberchrisparkXclairevodanshipperdemishassabisdingyidoodlesteinelithrarelonmuskericzakariassonhylarucoderimdigitalashishjakevin7jarredsumnerjerryjliu0joshalbrechtjpschroederkaiofreitaslexruslogancyangmitchellhpetradonkaprathamgrvprathyvshquanruzhuoxiuraroquerohanpaul_aisamasawyerhoodservasyy_aishao__mengsteipetetechgirl1908thorstenballtobiaswuptricalttuturetomuntraceable_thevelvet_sharkycombinatorziwenxu_
@dexhorthy·1d agoFor You
How to be good at ai coding, the parts orthogonal to the llm / ai skills and intuition
@dexhorthy
As @GeoffreyHuntley says “it’s just software engineering” ie some stuff is orthogonal to ai and all still true. Book recs include Clean code by @unclebobmartin pragmatic programmer, refactoring by @martinfowler - design patterns by GOF etc etc The other half is experience - as @nayshins says you know something’s bad because you’ve debugged it at 3am or spent weeks undoing some bad pattern
如何在 AI 编程上做好——那些与 LLM/AI 直觉正交的部分 【引用】正如 @GeoffreyHuntley 说的「这只是软件工程」——有些东西跟 AI 无关，依然成立。书单推荐：@unclebobmartin 的《Clean Code》、《Pragmatic Programmer》、@martinfowler 的《重构》、GOF 的《设计模式》等。 另一半是经验——正如 @nayshins 说的，你之所以知道某个东西有问题，是因为你曾经凌晨三点调试过它，或者花了好几周才把某个烂模式撤销掉。
@dexhorthy·3d agoFor You
I’m disappointed in anthropic but I am PISSED at the openclaw / Hermes grifters trying to steal inference that made this necessary
The market is the market so if something is possible and valuable people will find a way to get at it
Tbh probably should have seen this coming despite the “sdk is still fine” post from @bcherny a month ago
Cheap inference was a Weird blip in an increasingly weird world
我对 Anthropic 感到失望，但真正让我愤怒的是那些 openclaw / Hermes 的投机分子，他们试图蹭推理资源，正是他们让这一切变得必要 市场就是市场，只要某件事可行且有价值，人们就会想办法去做 说实话，尽管 @bcherny 一个月前发了「SDK 仍然可用」的帖子，但这件事其实早该预料到了 廉价推理，不过是一个日益怪异的世界里昙花一现的奇异现象
@dexhorthy·5月9日For You
Been helping users generate HtML to understand complex plans and changes for months now. But the real value of markdown is twofold, and html only hits the first one
1) clear, compact summary of a lot of information or intent - higher leverage faster easier understanding for HUMANS
2) a token efficient compact summary of information or intent for MODELS
it’s important to know what your goal is - more leverage for humans, or more performance from models/agents
If you need 2) then HTML is gonna blow up your context window way faster than 1)
We often find ourselves using both, markdown for most cases, supplemental html if there’s an opportunity to make content more digestible for humans
@trq212
HTML is the new markdown. I've stopped writing markdown files for almost everything and switched to using Claude Code to generate HTML for me. This is why.
几个月来我们一直在帮用户生成 HTML 来理解复杂的计划和变更。但 markdown 真正的价值是双重的，HTML 只命中了第一点： 1）对大量信息或意图做出清晰、紧凑的摘要——让人类更高效地理解 2）对信息或意图做出 token 高效的紧凑摘要——用于模型/agent 重要的是清楚你的目标：是给人类更多杠杆，还是从模型中获得更好的性能。 如果是目标 2），HTML 会比 markdown 更快地炸掉你的上下文窗口。 我们通常两者都用——大多数场景用 markdown，有机会让内容对人类更易消化时才辅助使用 HTML。 --- 引用推文 @trq212：HTML 是新的 markdown。我几乎停止为所有事情编写 markdown 文件，改用 Claude Code 为我生成 HTML。这就是原因。
@dexhorthy·4月6日For You
DO NOT ask a model to make a judgment call - they are trained to be sycophantic and to tell us what we wanna hear. The model has no idea if you've never taken a cs class or if you're linus f**** torvals. @vaibcode nailed it - you cannot outsource the thinking https://t.co/wUGQiZqwfc
不要让模型做判断性决策——它们被训练得具有讨好性，会告诉我们想听的话。模型根本不知道你是从没学过 CS 的新手，还是 Linus Torvalds。@vaibcode 说得很准：你不能把思考外包出去。
@dexhorthy·4月6日For You
feels like model providers have shifted from chasing 100% intelligence to some percentage split between raw intelligence and more-vertical products
i think its a signal that making models generally smarter is topping out and its more cost-effective to fill in the gaps with product (ie context engineering and task-specific RL - claude code, cowork, etc)
to me the split feels like 80/20 product/intelligence right now. Wdyt?
Is this because we have models capable of AGI and now we just need to wire them up?
Or is this just a symptom of hitting a wall and venture-backed businesses need to keep pumping value through other parts of the stack?
How long until the bitter lesson kills this? Year? Years? a decade?
感觉模型厂商已经从追求 100% 智能，转向在原始智能和更垂直的产品之间寻求某种比例分配。 我认为这是一个信号：让模型普遍变得更聪明正在触顶，用产品来填补差距（比如 context engineering 和任务特定的 RL——Claude Code、Cowork 等）变得更划算。 在我看来，目前这个比例大概是产品/智能 80/20。你怎么看？ 这是因为我们已经有了能达到 AGI 水平的模型，现在只需要把它们接上线？ 还是说这只是撞墙的症状，VC 支持的企业需要继续通过其他技术栈层来持续泵入价值？ Bitter Lesson 要多久才会干掉这种趋势？一年？几年？十年？
@dexhorthy·3月28日For You
lot of sandbox infra is building for the "cattle" use case but I think we'll see just as much use cases for "pets" as background agents become the norm.
I still find it hard to shift from "agents running in remote compute" to "agents running on my laptop" - not because of the agent or the inference, but because I'm used to having everything right in its place, libraries, clis, repo paths, etc etc.
The first step to remote coding agents should feel at least somewhat continuous in this regard
You bring the server or base image and configure it how you want with API keys, repo checkouts, etc - then the agent platform runs sessions on that host
sure you need some more plumbing to make it work in terms of compute efficiency but the cost of a small-to-medium EC2 is pennies compared to the cost of tokens I'm blasting through this thing.
enterprise side here interesting too...run everything in your VPC etc.
大量沙盒基础设施是为「cattle（牲畜）」场景构建的，但我认为随着后台 agent 成为常态，「pets（宠物）」场景会同样普遍。 我发现自己很难从「agent 运行在远程计算资源上」切换到「agent 运行在我的笔记本上」——不是因为 agent 本身或推理的问题，而是因为我习惯了把一切都放在固定位置：库、CLI 工具、repo 路径等等。 远程 coding agent 的第一步体验，在这方面应该保持足够的连续性： 你带来自己的服务器或基础镜像，按自己的方式配置好 API keys、repo checkout 等——然后让 agent 平台在这个宿主机上运行 session。 当然需要一些额外的管道工程来提升计算效率，但一台中小型 EC2 的成本，跟我砸进去的 token 费用比起来，简直九牛一毛。 企业侧也很有意思——把一切都跑在你自己的 VPC 里。
@dexhorthy·3月28日For You
there's gonna be an essay in ~6 months that resembles founder mode but it will be someone who got convinced by everyone to stop reading the code and then one day they woke up and everything sucked and they realized they needed to go back to "human mode"
大概六个月后会有一篇文章，类似「Founder Mode」，但主角是那些被所有人劝着不要再看代码的人——直到某天他们发现一切都变得一团糟，才意识到必须回归「human mode」。
@dexhorthy·3月27日For You
it’s a good post. I like to compare ai engineering to ci/cd - everyone wants to optimize it, it burns a lot of time, innovation is good but if everyone is innovating it’s chaos. Can you imagine if every third engineer had their own custom Jenkins server running under their desk?
@davidcrawshaw
No-one has figured out how an eng team should work with agents yet. Be wary of anyone telling you they know how to do it. Keep exploring. https://t.co/QZ3RXEyIzZ
这篇文章写得不错。我喜欢把 AI engineering 类比成 CI/CD——大家都想优化它，它吞噬大量时间，创新当然好，但如果每个人都在自己搞一套，就是混乱。你能想象每第三个工程师都在桌子底下跑一台自定义 Jenkins 服务器吗？ 【引用 @davidcrawshaw】：还没有人真正搞清楚工程团队该如何与 agent 协作。对任何声称自己知道怎么做的人保持警惕。持续探索。
@dexhorthy·3月24日For You
No Vibes Allowed - Living coding with Claude + CodeLayer: 🦄 ai that works https://t.co/1AXSn03K80
No Vibes Allowed —— 使用 Claude + CodeLayer 进行实时编码：🦄 能真正工作的 AI
357 tweets · 110 sources