懂车的可爱女孩小杏 @AliveUndead

YOU WON’T BELIEVE WHAT SPACEXAI ENGINEER LAUREN TAN SAID ABOUT RUNNING 100 AI AGENTS. “you can’t spawn 100 agents when you don’t even trust the output of one.” most developers think scaling AI means opening more agent windows. Lauren discovered the opposite. more agents only produce more work for you to inspect. if the agent cannot verify its own output, you become the verification system. it writes the code. you open the product. you find the failure. you capture the logs. you send everything back. then the agent tries again. you are not managing an AI team. you are manually closing every loop it leaves open. Lauren spent months building her way out of that loop. according to her: roughly 1,000 PRs shipped in one month almost 800 more by day 12 of the next around 20 PRs merged automatically while she slept the breakthrough wasn’t a smarter prompt. it was verification infrastructure. the agent could operate the real product. inspect traces. capture screenshots. read errors. test actual behavior. and keep working until the acceptance criteria were proven. then every repeated mistake became infrastructure: the same review comment appears three times? turn it into a lint rule. an agent keeps breaking an architectural boundary? make CI reject the import. the model always takes the easiest shortcut? redesign the environment until the easiest shortcut becomes the correct one. only then can one engineer safely move from: watch every output → verification → reusable skills → evals → auto-merge → agent teams the crazy part isn’t that Lauren can coordinate 10–20 Grok Bot agents. the crazy part is that she built a system where their work can continue without waiting for her to inspect every line. Grok Bot writes the change. the verification skill tests the behavior. CI enforces the architecture. the human reviews the evidence and controls the irreversible decisions. that is the difference between running multiple chatbots and operating an actual AI engineering team. more agents are not leverage. trusted verification is leverage. this free workshop is worth more than most $650 agent engineering courses. bookmark it and watch it today. then read how to build your own verified Grok Bot team below ↓

kozh ./ @mr_kozh

2 weeks ago

x.com/i/article/2109…

6 4 41 29K 22

@realsararafiz 对，这台价格就是照着折完的行情谈的，上一任车主把第一年的钱扛了，我算是捡了个现成的。你也看过同款二手吗？

呼和浩特这台二手升仕巡航像轻熟酷姐姐，车身有些使用感，低速却依旧好掌握。包头、鄂尔多斯、赤峰想看完整体验，主页有后续。

@MrsRiley09 先发照片吧，试骑感受得等我把鄂尔多斯那段跑完再一起写，你更想先看哪块？

@elreezy55 低速离合手感我觉得偏轻，市区走走停停捏着不累，不过每个人手劲不一样，你平时骑什么车？

@TINSLICE 低速就是原地挪车、掉头不用太抻，车把轻但不飘，我个头也一般，推起来不算费劲。你大概多高？我帮你对一下座高感受。

@viralismbec 接受这个说法，外观确实排后面。保养记录我这边还没细聊到，电瓶状态得现场看，你去主页看完整体验应该有数。

@zara1437 对，我也有这感觉，表面能挑出来的问题反而不慌，怕的是藏得深的。你之前那台后来出了什么幺蛾子吗？

@BreeBaLoo 包头那台低速掉头稳是稳，就是车漆难免有点小故事，你最后有没有真拿下？

@hapawoo 坐垫磨痕确实明显，但不是压着久放的那种塌陷，骑起来座感还撑得住。你也线下看过这台？

i gave Grok Bot 128 documents and told it one thing don’t summarize them then 2,774 nodes became 10,199 links, the raw files stayed untouched, and the bot started building a second brain around me while i was doing other work that’s what this video is showing sources stay locked new facts get pulled into separate pages projects connect to research decisions connect to context people connect to conversations and every useful relationship gets written back into the graph instead of dying inside one chat the part i care about most is that nothing has to “remember everything” the system only needs to know where the truth already lives by the end, 128 documents weren’t 128 documents anymore they were a knowledge engine i could keep adding to without rebuilding the context from zero i used to think AI memory meant giving a model a bigger context window now i think the better version is making sure useful knowledge has somewhere permanent to go

Lummox @Lummox_eth

a week ago

x.com/i/article/2111…

8 29 161 242K 409

@Asteri_eth i assume you have no clue sanity events is your own new excuse for death on camera speedrunning empty room full of chair real fact

沈阳今天碰到雅马哈仿赛，像成年大学校花穿上赛服，身形紧，车尾微微上扬，第一眼清冷。真正进入弯道以后却很会回应，方向给多少就跟多少。大连、鞍山、抚顺想看完整试驾，主页有后续。

@unicodef1wn So true

link: x.ai/bot/use-cases don't forget to save it

@irongiantXBT and the final room in this film a low fuss nightmare

SpaceXAI's Grok Bot team loaded their own templates into the marketplace and it's a f*cking goldmine you pick a setup made for a specific job, hook up your tools and mold it to your workflow here are 10 that deserve attention: bot 1 → dr eggbot creates other Grok Bots for you. describe the job you need handled and it assembles a specialist for it bot 2 → Lingxi's Engineer Bot watches over cloud coding agents and follows their PRs. you control the merge bot 3 → Stalk Bot tracks competitors and surfaces changes to their products, pricing and positioning - all with sources bot 4 → Haggle Bot catches wasted SaaS spend and drafts vendor counter-offers for you to check before sending bot 5 → Proto Bot turns feature ideas into working prototypes inside your app, delivered as draft PRs bot 6 → Video Editor selects the best takes from raw footage and puts together captioned vertical videos - edit plan goes to you first bot 7 → figma bro translates Figma frames into build specs and reviews your components and design tokens bot 8 → Distro Bot runs your marketing campaigns and creative - checks with you before spending, posting or sending bot 9 → Startup QA Bot navigates your product in a test account and documents bugs with screenshots and repro steps bot 10 → Signal Prospector extracts buying signals from your CRM and product data, then drafts outreach tied to the evidence worth reading through before you waste another evening building a system prompt from nothing

Iron Giant @irongiantXBT

2 weeks ago

x.com/i/article/2109…

53 64 497 838K 2K

xAI released 15 grok bot guides and almost nobody caught it each one written by the team behind grok the whole section is free grab a guide, follow it, ship your bot builders are still guessing when the instructions already exist 15 full guides, zero cost see them below 👇

Iron Giant @irongiantXBT

2 weeks ago

x.com/i/article/2109…

53 64 497 838K 2K

THIS IS F**KING INSANE She claims $69K/month from Shorts. She never starts from a blank page. Here’s the workflow: → Find kids channels already doing millions of views → Pick one viral video → Grab the transcript → Feed it into AI → Turn the structure into a prompt → Generate the scenes in Sora or Swaggi AI → Upload → Repeat Cute animated animals. Simple stories. Formats that already have attention. She’s not inventing a new concept. She’s studying what’s already working, extracting the structure, then making her own version with AI. Proven content → extract the structure → generate → publish → repeat Most people use AI to make one video. She’s using it to run a whole channel. Full workflow is in the video. Bookmark this and follow for more.

Isagii @isagi1000x

2 weeks ago

x.com/i/article/2110…

2 5 152 44K 68

x.com/i/article/2109…