Don't let AI write code blindly. In Antigravity, request an "Implementation Plan" artifact first. Review the architecture, edit the markdown logic, and then approve execution. Control the outcome, don't just hope for it.

And why we don't have DeepThink in Antigravity yet?

I got banned from 3 accounts from using your antigravity models 🙃

The Gemini Pro High and Flash models sometimes deviate from the plan and malfunction, but this feature works great on other models. I hope the Deep Think feature will be added to Antigravity as well.

Fix your quota issues on Pro accounts. It doesn't refresh every 5 hours..

Antigravity team, since you are started banning people for using your oauth key in other IDEs, i have a question. I don't care about other IDEs. I'm using oauth in my own version of Can i use it, that's not prohibited? I'm not using it with other IDE's i'm using in my own personal scripts. Its for multi research with multi agents in parallel work, very helpfull. It's better than one agent in hard sciencific task. (Since Gemini co-scientist is not avialable for regular users). If not - do something similar. Since Antigravity can use parallel agents so this is possible.

First fix your quota. Pro users get very limited quota then its a weekly wait. Its like reviewing the Implementation Plan for a week then trying to build something within 1 session

How does the "Implementation Plan" in Antigravity ensure adaptability to changing requirements? Does the review process involve team collaboration for refining architecture? 🤔

@antigravity I'm using the Gemini Pro plan, but right when development starts flowing smoothly, I hit the rate limits. It's a total buzzkill.

cool demo, now fix the ban, rate limits, and Gemini randomly deleting the code and we’ll talk

Totally agree. AI should draft, humans should design. Architecture first, execution second.

Don't buy pro plan on antigravity its basically an scam, they had 5hs refresh. Now they added an extra weekly limit but the limit for the pro plan is one window of 5hs and you get lockout...

At A2AX, we're working on adding skills from Antigravity and other accessible sources. Check out our new extension!

This is the difference between a Junior dev and an Architect. 🏗️🧠

The implementation plan is the best, especially because you can review it and comment on changes. That’s cool. It would be nice if, after a review, some of the user-expected changes were saved to the AGENTSmd file

Process over prompts. Structure first, then execution.

Your rate limits even for pro user are very strict. Please increase those.

Feedback - antigravity doesn't follow the instructions. While in plan mode, if you edit once, it begins implementation when it wasn't asked to begin. This wastes time and energy. I think it should wait for an explicit go ahead before implementation

I like to make really robust plans, then fast YOLO mode using flash model 😂

Antigravity is nice but you need to bring it to Enterprise customers (yesterday!) Gemini CLI is horrible but antigravity can win enterprise deals.

The strange thing is that the AI in the Agent Manager (compared to the editor) seems to be more advanced using the browser and implementation plans.

Planning before generation builds leverage. Architecture review catches hidden complexity. Execution after alignment saves time and refactors. Great builders design the path before they walk it.

All is fine but Gemini just deletes your codes for no reason .

Browser tool broke on my acc

What happened to the 5-hour limits? It used to be perfect for work, now I can't use it for anything. Before I get any useful results, the quota is reached and I have to wait several days to be able to use it again.

Curious—and why we don't have DeepThink in Antigravity yet?

Yes the task, planning, walkthrough and implementation features are impressive 👍

Arregle en mcp de navegador solo funciona en mac, en windows no funciona...

Or and just hear me out. Give it a solid plan and let it rip

Absolutely, actually most of us lost patience at least to do this!

Derp @GeminiApp got stuck trying to fix its own identation problem, burned down infinite tokens, now I cannot use any model.... @antigravity experience

upgrade limits first camon guys 20 dollar plan worth nothing gemini 3 pro is shit

I want magic 🪄. I want the AI 🤖 to know what I want and just do it. I don’t want to review. Even if I am unable to articulate my goals clearly. I want an AI that knows me like the addictive shorts know me. Lols 😂

Why doesn't the audio transcribe when using it? We might want to edit something and add some - say - error trace along with our spoken instructions.

Gemini in Antigravity is just the worse.