Unfortunately, Ai Studio Builder is unusable!
It used to be an excellent application, but it can’t even perform a simple task. You spend 20 minutes trying to implement something and can’t (until it times out and stops). I tested it on other devices and always got the same negative result.
Sadly, there were serious errors in this update attempt. They not only regressed, but they ruined the application - it became unusable (wasting time and yielding no results).
It’s strange, even in recent days, it’s true that the remixes are annoying, but for everything else it worked fine.
To be honest, the platform and the way apps stored update is a total shitshow. A week has passed already.
I’m out of polite words, typical Google. Do I understand correctly that there’s no longer a way to duplicate a project along with its chat history? I mean, my workflow lately (prior to these updates) looked like this: I make a copy of the project, write a prompt, check the result. If the result is bad, I delete the copy, open the original, duplicate it again, and submit the same prompt. The whole point is that if I had already made 10 requests in the chat, the copy would include that chat history. Meaning, the model remembers our entire dialog. So if I write a bad prompt, or it starts hallucinating and fixing things I didn’t ask for, I can simply roll back to the exact moment before I sent that new request. It was a 1-to-1 experience with a regular chat. You could revert to a previous point in the conversation at any time. NOT just reverting the code, but reverting it along with the chat history. That’s how it used to be. Now, the “Remix” button creates a “copy” of the project but completely wipes the chat clean. So, is there really no way to duplicate a project while keeping the chat now? If so, ‘thanks, Google.’
Don’t get me wrong, a button to reset the chat is also critically important. But so is the ability to fully duplicate the chat history. These are two critical features, and both of them should be available.
Also, from what I gather, instead of the simple chat we used to have, it’s now turned into some kind of IDE sandbox where the agent uses shell commands and other nonsense? This is just awful. I’ve tried Antigravity, and it doesn’t work very well either. The old, simple ‘chat’ worked infinitely better. And now it’s the exact same story here. For a straightforward prompt that used to yield a quick, high-quality response, I now just see a bunch of useless shell executions and files being shuffled around. What a nightmare.
Honestly, it feels like it would be easier to just get an API key and build my own custom UI for work, rather than banging my head against the wall with these AI Studio updates.
I am doing exactly this…since past 3 days…Honestly I am liking antigravity more than google ai studio now…
Do it! 
No regrets, though it’s probably their way of forcing us to AG
Hi,
I am writing to report a technical issue I’ve encountered while working on my project, “Robo Teacher AI”.
I am experiencing the following problems:
Preview Screen Error: The preview window is not loading and displays a “refused to connect” error (specifically for the europe-west2.run.app address).
Prompt Processing Issue: The chat interface is not processing my requests. Despite entering prompts for the teacher panel’s exam reporting section, the system remains stuck on “Unraveling” and does not generate any code or updates.
The build logs show that the connection is established, but the front-end rendering is failing. I have attached a screenshot of the interface and the console logs for your review.
Could you please investigate this issue and let me know how I can resolve it?
Best regards,
The quota has been exceeded. Please try again later. My membership is Ultra. Does anyone know when the problems will be resolved?
Hello… how do you able to view antigravity browser preview of your uplaoded app from AI Studio?
in ai studio we had preview… how can we have in this new Anti gravity
help please
Alguma notícia de quando vão corrigir?
Same for me.
I wonder if Google really found where their problems came from!
They decided to rollback and the problems remain.
For me they have a problem with system ressources access with paid requests…..
No land so far on the horizon Captain …
Sorry about the Ultra.
In terminal run npm install, then npm run build and then finally npm run dev
Providing you have 0 errors in your code this should give you a local host link, CTRL+left click this link and it will open in your default browser. 
Yes, I start with Anigravity too …and after 1/2 I´m on limitis with Gemini 3 flash …and new will be after 5 days …Google just kidding with us.
Nothing like that happened to me…It’s working great I use 3.1 Pro whole day - never got limit alert
same here i got banned for 5 days for using gemini models and we know what happens if you use claude
Honestly, it’s been almost a week and everything is still showing green on the status pages, but in reality the issues haven’t gone away. Things keep breaking in day-to-day usage.
I even tried Antigravity as a workaround and yeah, it helped a bit, but it’s frustrating that we have to rely on random extensions and hacks just to get basic things working again.
If anyone has found similar tools or alternative setups that make this workflow more stable right now, please share. At this point I’m more interested in something that actually works consistently than status pages that say everything is “fine.”
