Shen Zhuoran

Coding agents/self-improvement @xai. Ex-@GoogleAI Resident/@augmentcode. Alum @HKUniversity. 💎 Terran @StarCraft II. Views are personal.

[Shen Zhuoran @CMS_Flash]

In the internet age, people use apps for free in exchange of ads.

In the AI age, people can use physical services for free in exchange of AI training LOL.

[shift @joinshiftX]

Today, we're launching shift. We're starting by cleaning your apartment in New York City, for free.

Here's how it works. Book a shift cleaning. A vetted shift operator comes to your home wearing one of our devices. They clean. They leave. You pay nothing.

In exchange, we record the cleaning. Robotics is being built on data about how people do daily tasks, and the value of that recording is what funds the service. Anything personal in it is anonymized before the recording is processed.

By now, you have heard about the shift to AI more times than you can count. About the shift toward you, the part where you actually feel it, you have heard almost nothing. Shift is what starts to make it concrete, in specific cities, with specific services.

Today, cleaning in New York. Soon, handymen, repairs, and errands across the globe. And this is just one side of shift, with more on the way.

Comment “shift” and we’ll send you an early access link.

[Boyuan Zheng @boyuan__zheng]

Excited to see people try Grok Build for web dev. Our team has put a lot of effort into improving its aesthetics, functionality, and more exciting features to be expected with recursive self-improvement loop.

It’s still early beta, and feedback is very welcome. Please try it out and let us know where we can improve.

[Shen Zhuoran @CMS_Flash]

@boyuan__zheng @xai Finish our push for the best web dev model in the world 🔥🔥🔥!

[Shen Zhuoran @CMS_Flash]

I left @xai last week.

It was quite a journey. I learned a lot technically and made some lifelong friends.

I was fortunate to have spent almost all of my xAI tenure on a singular, long-term project on optimizing Grok's ability to one-shot complex, complete, and polished web apps.

I am proud of this team. We contributed one of xAI's most successful alignment recipes, which has been replicated to other teams. We scaled another innovative recipe to incredible scales, for which Grok 4.3 gave a tip-of-an-iceberg preview. I am excited to see its full impact reveal on future releases.

Looking forward, we are now at a critical point in the history of AI. Coding is on track to become a solved problem and AI is on the brink of its first total disruption of a major industry. Models show the first signs of substantial participation in their own development. The singularity might be on the horizon.

Today might be analogous to the eves of AlexNet and GPT-3, or perhaps more profound. I am excited for what is to come.

[Shen Zhuoran @CMS_Flash]

@JaySym_Ai How to get access to Cosmos?

[Shen Zhuoran @CMS_Flash]

@zhmeishi @xai It was a memorable journey pushing on web dev together with you!

[Shen Zhuoran @CMS_Flash]

@OpsoFacto @xai No.

[Shen Zhuoran @CMS_Flash]

@HoneybadgerFan4 @xai Sure, lemme know when you want to play!

[Boyuan Zheng @boyuan__zheng]

We put a lot of work into making it smarter and stronger at coding, especially for a small model.

Excited for what's coming next with bigger models training at Colossus 2.

[Shen Zhuoran @CMS_Flash]

@ar0cket1 Why?

[Shen Zhuoran @CMS_Flash]

1. Residual connections
2. Attention mechanism
3. Linear attention

[Core Automation @CoreAutoAI]

What are deep learning architecture modifications you don’t consider hacks @_arohan_

[Shen Zhuoran @CMS_Flash]

End-to-end eval is the way to scale to the future.

For general coding, it will be whole repo generation for production-scale libraries, tested by UTs.

For app development, it will be one-shotting production-grade apps, like Amazon, X, or Workday, tested by agentic grading.

[John Yang @jyangballin]

How much of SQLite, FFmpeg, PHP compiler can LMs code from scratch? Given just an executable and no starter code or internet access.

Introducing ProgramBench: 200 rigorous, whole-repo generation tasks where models design, build, and ship a working program end to end. 🧵

[Shen Zhuoran @CMS_Flash]

@Designarena @XiaomiMiMo @Xiaomi No scores yet?

[Shen Zhuoran @CMS_Flash]

Agents are pretty much just English compilers.

[skcd @skcd42]

2018: waiting for my compiler to complete

2026: waiting for my agent to complete

[Boyuan Zheng @boyuan__zheng]

Incredible to see what we've been cooking this past month — and this is only the beginning. Stronger models and richer capabilities are on the way.

What excites me most: what happens when the web dev model hits a singularity and unlocks Computer Use Agent self-play?

Proud to be part of the team building this🚀

[Beibin Li @beibin79]

Good job team!~ @ahmnav @CMS_Flash @arnogau @boyuan__zheng @ebecker_xai @1kevin_h @michaelbzhu @zhmeishi

[Shen Zhuoran @CMS_Flash]

Cool work of the team is finally out in the world.

This is a very early preview, and much more is to come around:

- One-shotting complex web apps;
- Pure vibe coding;
- Self-improvement between browser use and web app development.

The last point is critical and unique to web dev, because there is an intelligence gap between using a web app and building a web app that we can exploit, in theory leading to indefinite scaling of self-improvement.

[BijanBowen @Ominousind]

Grok 4.3 Beta browser os test result, gta clone and voice control apps

[Shen Zhuoran @CMS_Flash]

Coming soon!

[Elon Musk @elonmusk]

@HighlyUnspoken @cb_doge @Similarweb @grok Beta release of Grok Build app & terminal next week

[Nat McAleese @__nmca__]

A full-scale US Waymo rollout would cost ~700 full-time jobs in the funeral care industry (by saving around 35 thousand young American lives per year). Will no one think of (some of) the morticians!

[Shen Zhuoran @CMS_Flash]

@skcd42 Software -> game engine -> video gen, sounds about right