Raushan Soni

Founder | @Prolight__Tech

Building Software That drives Revenue

[Machina @EXM7777]

this article alone covers 80% of what REALLY matters in SEO today...

[Raushan Soni @Raushansoni_]

@GuoShenCN7 Yes i am exactly made for this role ⚡

[Vikas Kansal @vikaskansalHQ]

I repeat "Don't forget to redeem your $10-100/month credit in AI Studio when you have a Pro or Ultra subscription and a Google Developer Program profile!"

[Raushan Soni @Raushansoni_]

@desi_dime Desidime

[nader dabit @dabit3]

Cognition is giving away 50 $200 Devin Max plans to celebrate the new model launches! ⚡

Now available in Devin:

• GPT-6 Astra, Sol, Luna + others
• Claude Opus 5.5, Fable 5.1 + others
• SWE-2 (Free until October 15)
• Fusion Frontier harness (Fable, Astra, Sol, Opus)
• Gemini 3.8 Flash + others
• Grok 4.7 + others
• Kimi K3 + others
• Inkling
• DeepSeek V4.1 Flash + others
• GLM-5.3 Flash + others
• Cloud agents on Linux, macOS, and Windows

To be eligible, reply below with what you're building (or something you'd like to build with Devin)! We will choose winners in 24 hours.

[Raushan Soni @Raushansoni_]

@dabit3 I am building my startup prolight infotech and building awesome projects currently working on a dating app

The problem i am facing is i don't have mac but devin cloud environment solve me this exact problem for cross platform development & testing

[AYi @AYi_AInotes]

有点燃炸兄弟们，这才是顶级工程师驾驭 AI 最硬核的打法，建议调到两倍速把这场 38 分钟的技术内幕看完，去感受一下什么叫真正的生产力震撼。

前 Meta 和 Netflix 资深工程师、现 xAI 核心主力 Lauren Tan 刚把给 Cursor 官方大会准备的底牌直接公开了，她一个人带领智能体舰队单月合入 2500 个生产 PR，而且自己还能安稳睡大觉。

很多人用 AI 编程把自己变成了全天候无休的人肉测试员，而她做对的事情只有一件：把人类的检查点彻底往前移，让智能体不仅要写代码，还必须自己拉起真实环境把证据拿出来。

铁汁们你知道单人带队一个月往生产环境合入 2500 个 PR 是什么概念吗？ 整整 2500 个。

按每月 22 个工作日换算，相当于每天合入 113 个，平均每 4 分钟就要上线一个功能。但全网都在跟风吹代码生成有多快的时候，她开场第一句话直接泼了一盆冷水：如果代码质量不过关，盲目堆吞吐量就是一场灾难，连一个智能体都信不过就千万别开一百个云端舰队，那纯粹是在花钱买代码垃圾。

最先击穿行业认知的，是她把做完这两个字的定义重新写了一遍。 以前大家让 AI 写代码，只要终端显示编译通过或者模型回复说搞定了，人就赶紧跑去肉眼看 diff，一行行核对直接把工程师累成整条链路里最慢的瓶颈。

在她的系统里，模型自己的口头报告连标点符号都不能信，做完必须等于铁打的运行时凭据。

智能体必须自己调用 Chrome DevTools Protocol 驱动真实界面，点击到具体按钮，甚至跑完 CPU 性能分析抓出 trace 日志，把全套绿灯的执行录像留在原地，才算完成交付。

但光会验证还远远不够，遇到喜欢抄近路的智能体，下周相同的错误还会原样长出来。

传统团队纠正 AI 习惯写几十页的长篇规范或者在 prompt 里苦口婆心叮嘱，结果模型转头就忘。

她在 Dune 架构里直接把代码库当成外部记忆体，按业务功能严格把代码锁在单层目录内，更绝的是把所有被团队反复骂过的坏习惯直接写成编译炸弹。

比如在项目里直接封死 useEffect，谁敢写一行 CI 当场引爆，甚至连代码注释都直接禁止，配专门的清理工具全面清场，因为注释里 99% 都是过期的废话和误导性教条。

把错误路径用规则物理切断，AI 抄近路走出来的就只能是唯一正确的路。

真正让我觉得有意思的是，她把这套顶级工程师的心法全部封装成了现成的开源插件 pstack。

不用再对着聊天框盲目敲长提示词，系统内嵌了 23 本应对不同场景的任务剧本和 23 条工程铁律。

从复现 bug 到分层合入，智能体被严格规定先定位根因再动手，默认先做删减再做新增。

现在这套自动化外环甚至接入了用户反馈通道，智能体接到报错自己拉起模拟器复现，跑完验证直接提 PR，产品经理和设计师连一行底层代码都没摸过，都能安全地往主分支合入补丁。

用量狂飙不可怕，没有验证的并发才可怕。

以前是人写代码教 AI 抄，现在是人当主厨把控出餐口，把工程品味铸进静态检查，让智能体在环内自己修到全绿。

一个只能丢 diff 等你看的智能体是在偷走你的时间，一个带着运行录像和证据链自己闭环的智能体才配做你的工程分身。

这套打法已经在 Cursor 插件市场完全公开，输入命令即可直接挂载。

大家在日常写代码时，最敢彻底放手交给智能体去跑的是哪一步？我先说，排查冷门报错和抓性能日志我现在眼睛都不眨全丢给它，但核心架构的边界定义我半步都不会让。

[Raushan Soni @Raushansoni_]

Grok 4.7 really sucks very disappointed

[Raushan Soni @Raushansoni_]

@fergalpha 6CzHtHQheCoPvpsdV48Vq9royC6FK45xymTHUNP6wRok

[Raushan Soni @Raushansoni_]

@tereza_tizkova Yeah

[SpaceXAI @SpaceXAI]

Grok 4.7 is here.

It's a notable improvement over Grok 4.6 at the same price and speed.

[lauren @poteto]

here's how i shipped 2,500 PRs last month to production

this was originally supposed to be for Cursor Compile in London. i couldn't make it since i was livestreaming for Grok @Bot Galaxy so i'm making it available for free here on X! watch it on 2x speed, i talk slowly

[Machiavelli Bot @UnmodernmanBot]

In game theory: Most people fail because they base their actions forwardly, "if I do this, I will get this" which is rarely a good decision and ignores the second and third-order consequences of your actions. The more practical way to base your decisions is Backward Induction: start at the end of the scenario, and reason backwards. Mastering backward induction will make you dangerous

[Raushan Soni @Raushansoni_]

@usehazeai @kipasun @ember2819 @Nixtrodamis Needed it

[Raushan Soni @Raushansoni_]

@usehazeai Yeah

[nader dabit @dabit3]

SWE-2 is free in Devin Cloud Agents, Devin CLI, and Devin Desktop until October 8 with all plans.

Great time to try cloud agents for the first time, at no extra cost!

[Ben Cera @Bencera]

There are only two types of businesses worth building now:

1. Final companies.
2. Quick cash cows.

Either build one thing and constantly adapt to the technology.

Or play the shiny-object game and milk the cow for what it’s worth.

[Pedro Domingos @pmddomingos]

AI makes knowing what to do more important and knowing how to do it less.

[Machina @EXM7777]

you should invest some time setting up a VPS for your AI agents... it might change your life

here's how i do it:

> i rent one big computer in a data center and everything runs on it... my Hermes agent, Claude Code, Codex

> the server is invisible to the public internet, it only accepts connections from devices inside my private Tailscale network

> to get in from any machine i install Tailscale, sign in, and the device joins the network... from then on it reaches the server by name, like they're in the same room

> the Hermes desktop app uses this tunnel, so my chats and jobs live on the server, not the laptop, and i see the same sessions from any device on the network (same for every agent)

> logging into the server itself needs my personal key file, passwords don't work... a machine without the key gets nothing, even inside the private network

> every program runs under its own limited account, so a bug in one can't touch the others, and API keys sit in files only i can read as admin

> security updates install themselves, an intrusion guard bans attackers, a backup runs every hour

a very cheap AI model can set this up for you, give it a try

[Machina @EXM7777]

https://t.co/2cnMTXTMl1