Skip to main content
Andrea Verlicchi

Making the web faster and more user-friendly

Google Antigravity 2 & Gemini 3.5 Flash: this AI duo changed my dev workflow... again

It has only been a few days since I started using it, but I’m already incredibly enthusiastic. The speed, the precision, the planning process, and the final results obtained are simply outstanding.

The Google Antigravity logo featuring the text 'Experience liftoff'.
Google Antigravity 2 is here, and it's a complete game-changer.

Let me take a step back. Antigravity 1 was essentially a VS Code shell with a built-in chat. It felt quite similar to the VS Code + Copilot setup, but with a highly interesting twist: Planning Mode. Before executing any changes, the chat would pause, reflect back what it understood, show you a detailed step-by-step implementation plan, and ask for explicit confirmation. It was very similar to Claude's planning mode but packaged in a much cleaner, easier-to-read, and commentable user interface.

With Antigravity 2, they’ve taken a bold step. They’ve completely removed the code editor mode (or they’ve hidden it so well that I haven't found it yet!). Instead, it is a pure, distraction-free chat interface focused entirely on the agentic workflow, while keeping that wonderfully convenient Planning Mode from version 1 completely intact.

Seamless Multi-Project Workflows #

What makes Antigravity 2 so powerful is how it operates in the background:

I recommend using it side-by-side with your favorite traditional IDE. In my day-to-day work, I constantly jump between Antigravity 2 and JetBrains PyCharm. I use Antigravity to run agent tasks, plan refactoring, and modify code, and then I switch over to PyCharm to run, test, and handle everything else just like I did before. It’s the perfect hybrid workspace.

Under the Hood: Gemini 3.5 Flash is Fire #

Antigravity 2 supports both Gemini 3.5 and Claude. If you have a Google account, you get a generous free tier of Gemini 3.5 included right out of the box, which I highly recommend trying. If you prefer Claude, you can simply connect your Anthropic account and go.

I’ve personally tested almost every major model out there—from OpenAI's o3, to Gemini 2.5 Pro, to Claude Sonnet 3.7 and Opus 4.6. Over time, I’ve consistently found myself appreciating the quality and relevance of Gemini's answers the most.

And Gemini 3.5 Flash? It is absolute fire (or as we say in Italy, una bomba). It is blisteringly fast, incredibly accurate, and highly intelligent. You really have to try it yourself to understand the speed difference.

What Are You Waiting For? #

By the time you've finished reading this paragraph, you could have already downloaded it. And in the time it would take you to write a reply to this post, you could write your first prompt and experience it firsthand.

If you are a developer looking to supercharge your productivity, give the Antigravity 2 and Gemini 3.5 Flash duo a spin. I think you won't look back. 😊