Antigravity and Jules

My original motivation for subscribing to Gemini AI Pro was to access the advanced chat model and the Google Workspace assistance features. However, since the plan also includes usage quotas for Jules and Antigravity, I started experimenting with them to see how they fit into my development process.

For my daily tasks, these tools have settled into specific niches that improve my workflow, distinct from other web-based coding agents I have used in the past.

Previously, I used the web interface for Codex. While powerful, I found the web UI to be incredibly heavy, often causing Google Chrome to hang or enter a “waiting” state. It created friction in what should be a quick process.

In contrast, Jules runs comfortably. The interface is snappy and responsive, which is critical when I want to iterate quickly. It isn’t perfect—it occasionally gets stuck in an infinite loop executing the same command—but for routine tasks like small bug fixes or package updates, it handles the job without the browser lag.

The standout feature for me, however, is Antigravity, specifically its “Agent Manager” mode.

Most chat-based AI tools require immediate, synchronous interaction. You type, you wait, you reply. It feels like a conversation that demands your full attention. Antigravity feels different; it is closer to sending an email to a colleague to handle a task in the background.

I can fire off a request and let it run parallel to my own work. I particularly appreciate two features:

1. Inline Comments on Plans: I can review the agent’s proposed plan and add comments directly in context before execution.
2. Task Queuing: This is the most significant UX improvement. If I think of a task while doing something else, I can explicitly queue it for later execution. It doesn’t force a context switch or interrupt my current “field work.”

Having these included in my existing subscription has been a welcome addition to my toolkit.