My Hands-on Experience with Antigravity 2.0: Building a Todo App with Google Sheets as a Backend.
I recently spent some time testing Antigravity 2.0 through a small hands-on project.
Since Cursor has been my primary AI coding environment, I wanted to understand how Antigravity’s workflow feels in a real development scenario. To make the test practical, I built a simple Todo App step by step within approximately 30 minutes, using Google Sheets as the backend.
The goal was not to determine which tool is better. Instead, I wanted to observe how Antigravity approaches development compared to a workflow I am already familiar with.
A few key points stood out during the experiment:
- Antigravity follows an agent-first workflow, where the AI plans before writing code.
- Antigravity Desktop and Antigravity IDE serve different purposes, which may be confusing at first for users coming from other AI IDEs.
- Artifacts are useful for reviewing the AI’s reasoning and implementation plan before moving deeper into the actual code.
- Browser Subagent currently has some Windows-specific limitations, which are worth understanding before using it in a browser-based development workflow.
I documented the process based on actual usage, including screenshots, workflow observations, and an initial prompt example that others can adapt for their own experiments.
This article may be useful for developers, AI tool users, or anyone currently evaluating whether Antigravity 2.0 fits their development workflow—especially those coming from Cursor or other AI-assisted IDEs.
The article also includes real usage screenshots and a basic prompt example that can be adapted for your own projects.
Read the full article here:
Top comments (0)