Task automation is apparently one of those “best of Gemini” features. It’s already on some recent Pixel and Samsung Galaxy phones, and it enables Gemini to use certain apps on your behalf. It’s been limited to a handful of rideshare and food delivery apps until now. That’s changing “soon,” says Google, when task automation will open up to a wider range of apps.

It will also add multimodality; previously, Gemini could only use voice or text prompts to inform its actions. Now you can throw a screengrab or a photo into the mix, which kind of seems like something you should have been able to do from the start. You’ll be able to give Gemini a screenshot of a grocery list in your notes app and it will add those items to your cart. You know, provided you have an Android phone that supports Gemini Intelligence.

Another feature under the Gemini Intelligence umbrella that’s all new is Create My Widget. Google’s blog post calls it a first step toward “generative UI,” and it allows you to describe the functionality you want in natural language and let AI create a custom widget. Google’s examples include a custom weather widget for a cyclist who wants to see wind speed and precipitation at a glance, and a dashboard to surface specific recipe suggestions, like “three high-protein meal prep recipes every week.” These widgets will also carry over to Wear OS, so they’ll be available on watches, too.

It’s a pretty simple idea on the face of it, but if you think about widgets as tiny apps you can vibe-code right onto your phone’s homescreen, it gets a little more interesting. Maybe this really is a small step toward an interface that just creates itself on the fly. Or maybe that’s a lot of pressure to put on a humble widget feature. Either way, I’ll be curious if we hear more during the I/O keynote about “generative UI.”