AI Pulse · Under the Radar

A small project called moli showed up on GitHub this week and it’s worth a second look if your team touches web automation or AI agents at all. Most “browser automation” tools right now are wrappers around Chromium, they drive a real browser through a script. Moli is different. It’s a browser engine built from scratch in Rust, with no Chromium underneath it, designed specifically for AI agents to control through standard interfaces like CDP and WebDriver.

The interesting part is what it leaves out. Moli treats expensive rendering work, like layout and painting, as opt-in only. Normal browsers do that work constantly because a human is looking at the screen. An AI agent usually isn’t looking at anything, it just needs the page structure and data. By skipping the visual rendering unless it’s specifically asked for, moli can run faster and lighter than a full browser, which matters a lot if you’re running hundreds or thousands of agent sessions instead of one person clicking around.

It picked up 461 stars in its first week, which is fast for a niche infrastructure project, and it’s backed by a company that already runs cloud browser infrastructure, so this isn’t a hobby experiment with no runway behind it. Details on the project itself are still limited since it’s brand new, and how it performs at real scale is unverified so far.

For most business owners this won’t touch your day to day yet. But if you use or plan to use AI agents that browse the web, fill forms, or pull data from sites on your behalf, the tooling underneath them is about to get faster and cheaper, and that’s exactly the kind of infrastructure shift we track when we think about what should sit inside an AI command centre like the one we build with Omni.