Pinned Tweet
我要推广一个Mac 上最好用的图片查看和对比软件:
FrameSight 官网: framesight.ronn.fun/index_zh
App Store 🔗:apps.apple.com/app/framesigh…
今年 3 月上架,半年的自然流量还不够开发者账号的成本,接下来要好好打打广告,好酒也怕巷子深啊。
1
5
369
Ronny|AI影像 × 独立产品 retweeted
Winner announcement in 24 hours! We'll randomly draw 100 eligible winners with help from @grok. Make sure you follow us, like & repost to be in the running. Good luck! 🍀
Ronny|AI影像 × 独立产品 retweeted
Replying to @Jamesmv49
看到很多人评论不如这个不如那个的,大家要理解这个设备的定位:开箱即用的 openwrt 便携路由器!
论性能不如最强的,论价格不如最便宜的,但就是有它的受众群体啊。很多不愿意折腾的人,图的就是一个省心,或者说愿意支付溢价换一个省心。
我举个例子:理光 GR 系列的相机,现在卖接近 1w 为啥还那么多人趋之若鹜,就是因为那个定位独一档。没有可打的。
说了这么多可能会有人认为我在杠,但经过我的研究还真给大家找到了一个平替方案:中兴 U30Air 刷 MU300 Linux 固件,这样你就花不到 400RMB 拥有了一台具备 openwrt 带电池的随身路由,并且重量才 150g。大家拿好不要谢
Agent.Space 推出了Mac mini 租赁服务,$160/月,包年$1440,并且产品介绍页写的就是给Claude Code/Codex 等用的,听起来是个很不错的生意,但是有没有其它服务商把价格打下来啊。
@dingyi 我就是之前看你推荐,一直在关注它们家什么时候能推出 claude+codex 的组合订阅包,结果过了一个月还是没有 claude 订阅套餐。
1
58
原来我不是一个人,我还以为是我技术深度不够,看不懂 AI 输出的技术方案名词😂
但我觉得最大的问题不是输出看不懂,而是模型能不能理解我要干什么?它输出的内容能不能匹配我的决策方向。
所以我在在 AGENTS.md 上加了一层新的约束:把 Agent 的复杂分析和执行结果,转换成人可以快速理解、判断和决策的内容。
以前 Agent 经常直接把分析过程、技术细节、多个方案、文件改动和测试结果一次性全部输出给我。信息很多,但我经常看完之后仍然不知道:这一轮到底在解决什么问题、现在需要我做什么决定、这个决定会影响什么。
现在我要求 Agent 默认按照固定顺序输出:
问题 → 方案 → 推荐 → 为什么 → 影响 → 下一步动作,同时采用ASD-STE100 的表达方式:短句、一个句子只表达一个主要意思、统一术语、尽量使用主动语态和具体描述。
我只需要看到足够做判断的信息,只有需要深入时才继续展开实现细节。这样做的目的不是让 Agent “少说话”,而是把 Agent 的输出从执行报告变成真正适合人能看懂的决策信息。
如果你用 codex / claude code 落实文档驱动开发,你肯定会被技术文档搞的晕头转向。
这不是你的问题,现在 AI 产出的技术文档就是多变且难读。
海哥已严肃学习卡帕西 @karpathy 推荐的 ASD-STE100 技术文档写法,做了一个 hai-simplified-tenchnical - 一个专门用来规范/重写文档的 skill,大幅提升可读性,虽然篇幅有所增长,但每次只讲述一个点,没有太多理解上的障碍,读起来反而速度特别快。
具体的比对案例可以看评论区。
开源地址 github.com/hylarucoder/hai-s…
57
Ronny|AI影像 × 独立产品 retweeted
Any open source builders who haven't tried Devin yet? 🤔
Let's support the open source fam!
5 Devin Max plans to give away
Only rule:
Drop a comment - What open source project are you working on + public github repo link.
Winners announced:
Monday 5th Oct 8PM CET
🤖 Made with AI
Ronny|AI影像 × 独立产品 retweeted
I’ve got 10 FREE Devin Max plans ($200 each) to give away 👀
Y’all have seen me talk about and use Devin for my builds. If you’ve been wanting to try it, and figure out what the hype is all about, this is your chance.
What would you build?
Drop it below. I’ll pick 10 people and DM you a code. Let’s build!!
作为一个勉强懂点儿技术的产品经理,我对这一点感受尤其深刻:理解 LLM 输出的技术方案太晦涩难懂了,经常需要对它给出的方案仔细确认影响范围,收益价值。
AK 文章中其实还隐藏了很多观点:
1. 理解正在成为比生成更稀缺的资源。Agent 自主干活越多,人类更看不懂机器在干嘛。
2. 当前的模型远比你想象的要强大:大多数人眼中低估了当前模型能输出内容的丰富性,且还在按以前 chatbot 的习惯去约束和限制模型的发挥。
3. 软件和代码已经变得廉价,稀缺只剩下品味和判断。这对 PM 来说是好事,定义“什么是好的”这件事,短期内还无法被 LLM 替代。
We'll be spending a lot more time trying to understand the outputs of language models. A few thoughts, tips & tricks:
Writing. Something I've had success with: Ask your LLM to explain something in ASD-STE100, it's a controlled language specification originally developed for aerospace maintenance documentation. LLMs well-versed in this language and it comes with heavy constraints on clean writing style that I often find a lot more readable. Sometimes I've tried to soften it a bit e.g. ask for "80% of the way to ASD-STE100" because the spec is quite stringent. But even better:
Diagrams / images. Instead of writing, ask your LLM to create a diagram. These can be a lot easier to process, parse, and understand. But even better:
Web pages. Ask for output "in HTML" to get a beautiful, interactive webpage. LLMs are getting really good at frontend and can create beautiful experiences, animations, etc. But even better:
Explainer videos. The output format I am most bullish on is fully custom / bespoke explainer videos generated on any arbitrary topic. Experiment with things like "Create a 3b1b style video explainer on X. Use my ElevenLabs API key for audio narration". (you'd need an API key for the latter or you can ask your LLM to find you decent free alternatives that use your local compute). This is actually starting to work!
In summary:
- As LLMs get better, they will do more and more of the legwork autonomously, and a lot more of our work will rise up the abstractions into oversight and understanding.
- Luckily, LLMs can help here too because as intelligence and code are increasingly abundant, you can ask for large, custom, discardable software artifacts (e.g. web apps, video explainers) that would have never made sense to create before. Push the boundaries here and you'll be surprised.
43
Ronny|AI影像 × 独立产品 retweeted
We'll be spending a lot more time trying to understand the outputs of language models. A few thoughts, tips & tricks:
Writing. Something I've had success with: Ask your LLM to explain something in ASD-STE100, it's a controlled language specification originally developed for aerospace maintenance documentation. LLMs well-versed in this language and it comes with heavy constraints on clean writing style that I often find a lot more readable. Sometimes I've tried to soften it a bit e.g. ask for "80% of the way to ASD-STE100" because the spec is quite stringent. But even better:
Diagrams / images. Instead of writing, ask your LLM to create a diagram. These can be a lot easier to process, parse, and understand. But even better:
Web pages. Ask for output "in HTML" to get a beautiful, interactive webpage. LLMs are getting really good at frontend and can create beautiful experiences, animations, etc. But even better:
Explainer videos. The output format I am most bullish on is fully custom / bespoke explainer videos generated on any arbitrary topic. Experiment with things like "Create a 3b1b style video explainer on X. Use my ElevenLabs API key for audio narration". (you'd need an API key for the latter or you can ask your LLM to find you decent free alternatives that use your local compute). This is actually starting to work!
In summary:
- As LLMs get better, they will do more and more of the legwork autonomously, and a lot more of our work will rise up the abstractions into oversight and understanding.
- Luckily, LLMs can help here too because as intelligence and code are increasingly abundant, you can ask for large, custom, discardable software artifacts (e.g. web apps, video explainers) that would have never made sense to create before. Push the boundaries here and you'll be surprised.
最近@cognition 产品的迭代速度真的很快。
这不,Devin 手机 App 的内测版就可以体验了,跟我预想的有点儿不一样,我最开始以为也是像 Codex/Claude 那样去控制自己的本地电脑,体验了一番之后发现 Devin 走了另外一条截然不同的路:你可以自己配置 Ubuntu/macOS/Windows,也就是说你随时随地拥有了一台在云端的服务器,可以把自己的代码跑在属于自己的服务器上。
我觉得现阶段相比其他家最大的优势就是有 macOS 和 Windows,你可以把当前在家里或者公司的环境完整的搬到云服务器上。
真的是随时随地都可以 VibeCoding 了
#devin
35
我理想中的 Memoh 场景,是拥有一个真正持续在线的 AI 开发环境。
在家时,我会用 Claude Code、Codex 开发 LensMind 和 FrameSight;出门或者到了公司后,我希望只需要连上自己的云端环境,就能继续访问同一个项目、上下文和工具,在不同 Agent 之间继续开发。
不用再关心 Session 在哪台电脑、上一个 Agent 做到哪里,也不用每换一台设备就重新恢复环境。
我想要的其实很简单:开发环境跟着我走,而不是我被开发设备绑住。
如果 Memoh 最终能把 Cloud Computer、workspace、memory 和多个 coding agents 真正串起来,这就是我最想要的一整套 workflow。
还有一个我很期待的功能:如果未来也能像 Devin Cloud 一样提供完整的 macOS 云端环境就更好了。 对 iOS / macOS 开发来说,这意味着 Xcode 编译、模拟器调试、打包签名和测试环境也能一起常驻云端,真正做到人在任何地方,都可以接着让多个 Agent 继续开发同一个产品。@memoh_ai
We want to hear how you actually use Memoh.
Quote this post with your story:
the task you handed off, the workflow you stopped doing by hand, the moment it clicked.
We'll provide 3 Pro Membership Giveaway— $60 a month, on us.
32
最近真的被你们圈粉成 devin 推荐官了@cognition
1
29
Ronny|AI影像 × 独立产品 retweeted
Devin Mobile is now available in beta.
Join the waitlist at the link below.
































