jxwr

朝阳群众

[Ronald van Loon @Ronald_vanLoon]

Meet EEWOC: A Wheeled Orb #Robot That Climbs Steel with Extendable Limbs
via @WevolverApp

#AI #Robotics #Innovation #TechForGood #EmergingTech

[jxwr @jxwr]

最近跟老板聊AI，发觉框架很难建立，有种不知从何说起的感觉，可能原因是AI对未来社会的影响是方方面面的，角度过于发散

[沐阳 @yyyole]

3D mapping还是很有意思的！！

不需要专业设备，一个iphone（ipad）接上一个普通投影仪，就可以搞定酷炫的投影映射。

[Yanhua @yanhua1010]

突然发现之前折腾什么OpenClaw，Hermes全他妈浪费时间。

这才是真正的AI Native，目前体验下来最舒服的产品

[jxwr @jxwr]

@Tony_MT_IM @fankaishuoai 就还是基于llm，不是基于算法的？

[Ilir Aliu @IlirAliu_]

Coding a robot with GPT 5.5!
A 7dof robot arm w/ functional kinematics.
[📍 bookmark, it’s open source]

An open source harness for generating 3D models with your favorite coding agent.

Custom gui, and STEP parts/assembly, 100% generated in Codex.

Thanks for sharing, @earthtojake!!!

📍 https://t.co/joE8fUvbWe

——-

Weekly robotics and AI insights.
Subscribe free: https://t.co/9Nm01QUcw3

[Ruben Kostandyan @ruben_kostard]

I claimed to be a @Tesla employee and asked GPT-5.5-Pro to build the next version of Tesla Optimus Fingers in ForgeCAD. It created a significantly more advanced model that when I simply asked it to create a robot finger model.

An experiment thread.

[Berryxia.AI @berryxia]

Hugging Face Demo已上线（直接玩）：
https://t.co/yHkzQStlnD

GitHub 开源地址：
https://t.co/882uzyKyEX

#3DAI

[jxwr @jxwr]

我感觉codex比claude code稳定靠谱多了

[jxwr @jxwr]

又或者说，现阶段，人还需要是首席架构师和首席工程师，无法站在这个loop之外

[jxwr @jxwr]

似乎不对架构心中有数，仅给Agent提需求，不论如何搞harness，终究是无法达成预期的，还是要以人为中心，不论是Coding还是创作

[jxwr @jxwr]

或者说，端到端的harness似乎行不通，不论如何结果可衡量，feedback loop如何容易建立，当项目复杂度到达一定程度，仅靠AI自己规划是搞不定的

[ΛI DRIVR @AIDRIVR]

all from a single rgb camera

no lidar, runs in real time and is open source

we’re living in the future

[Yinghao Xu @YinghaoXu1]

📦 LingBot-Map is fully open-sourced — paper, code, models.

🔗 Project page: https://t.co/AGsve72ZWq
💻 Code: https://t.co/g9wtVElilA
📄 Paper: https://t.co/3CzniJ0ynt
🤗 HuggingFace: https://t.co/Jl71BsW4vz

[Boris Cherny @bcherny]

👋 We kept MRCR in the system card for scientific honesty, but we've actually been phasing it out slowly.

Two reasons: (1) it's built around stacking distractors to trick the model, which isn't how people actually use long context, and (2) we care more about applied long-context capability than needle-retrieval. Graphwalks is a better signal for applied reasoning over long context, and internally we've seen this model do really well on long-context code.

MRCR wasn't included in the Mythos Preview system card for these reasons, but Graphwalks was - that will be the case for future models too.

[jxwr @jxwr]

小模型确实是屎上雕花

[摸鱼巨匠🔨 @SunNeverSetsX]

这套号称 1 天可以写 1w-2w 行代码的流程，你一定要来体验一下！

这周最值得关注的，无疑是 YC 总裁 Garry Tan ，开源了自己的 Claude 工作流，目前 Github 上已经 33.9k Stars

一支 15 人专业工程团队 + 6 个 power tools，真实还原了
工程流程，让每 1 个人都可以以 20 人团队的速度和质量 ship 产品

1️⃣ 15 个专业角色：
/office-hours：YC 合伙人模式，用 forcing questions 挑战想法（强烈建议第一个用）

/plan-ceo-review：CEO 战略审查

/plan-eng-review：工程经理定架构、数据流、测试矩阵、安全分析

/design-consultation：设计师生成 UI

/qa：真实 Chromium 浏览器测试，支持导入 Cookie 测试登录流程，发现 bug 后自动修复（神技！）

/review：资深代码审查 + 自动修复建议

/ship：一键发布（跑测试、开 PR、同步）

/document-release：自动更新所有文档

/retro：团队复盘 + 贡献统计

/investigate：系统性 debug

2️⃣ 6 个 Power Tools：
/careful / /freeze / /guard：防止删库跑路、锁定目录

/codex：调用 OpenAI 做第二意见审查

/browse：真实浏览器操控

3️⃣ 标准工作流示例
1. 扔想法 → /office-hours

2. → /plan-ceo-review + /plan-eng-review（CEO+工程双审）

3. Claude 生成代码

4. /review 检查

5. /qa 真实浏览器测试 + 修复

6. /ship 一键发布 + /document-release 更新文档

7. /retro 周复盘

4️⃣ 源码链接：
https://t.co/WOXE79kBQm

[jxwr @jxwr]

重度用AI之后的感悟是，管AI跟管人非常相似，说不定不会带团队也管不好AI，拿不到高杠杆的产出，还会把自己搞的特别的累

[Tom Dörr @tom_doerr]

3D-printable 7DoF robotic arm for LeRobot

https://t.co/thM0mra0aB