You’re staring at a product page with six nearly identical laptops, trying to explain to a chatbot which two you want compared. “The second one from the left — no, the one with the silver bezel” — it’s like giving directions to someone who refuses to look out the window. Chrome 149 fixes this. Google’s new “Select from screen” tool lets you drag a box around anything on your current tab — text, images, product cards — and feed it directly to Gemini’s side panel. No copying. No describing. Just point and ask. If you’re exploring other AI-Powered Websites that can boost your workflow, there are plenty worth knowing about.
How “Select from Screen” Actually Works
It works like Circle to Search on Android — same concept, translated from touchscreen gesture to desktop pointer.
The mechanics are deliberately simple, which is exactly the point. Google’s own Chrome developer blog frames the feature as saving users from “having to describe exactly what you mean.” You select a region visually; Gemini responds to what you actually highlighted. Here’s the quick-start:
- Update to Chrome 149, then restart your browser — if you run into any setup snags, it helps to know the most common computer problems and how to fix them
- Click the Gemini icon at the top of Chrome to open the side panel
- Hit the “+” menu and choose “Select from screen”
- Drag a selection box over any text or image region on your current tab
- Press Esc to cancel, or resize your selection before sending
Available on macOS, Windows, and Chromebook Plus. The side panel runs on Gemini 3; computer-use and developer capabilities run on Gemini 3.5 Flash.
The Bigger Ambition Behind the Button
Auto Browse, Connected Apps, and a new commerce protocol round out the update.
“Select from screen” is one piece of a much larger Gemini-in-Chrome overhaul. Google has pinned Gemini’s side panel permanently to the browser, wired it into Gmail, Calendar, Shopping, and Flights through opt-in Connected Apps, and launched auto browse for AI Pro and Ultra subscribers in the U.S. — turning Chrome into something closer to a Spotify Wrapped for your intentions, anticipating what you need before you finish typing. Auto browse fills forms, logs into sites, and makes purchases on your behalf, though it pauses for confirmation before any sensitive action. Shopify, Etsy, Wayfair, and Target have co-developed a Universal Commerce Protocol to make agent-driven shopping work seamlessly across retailers.
The Part Worth Thinking About
Useful and worth scrutinizing aren’t mutually exclusive — and Google’s own controls reflect that tension.
Most AI browser features stop at answering questions. Auto browse and Gemini 3.5 Flash’s computer-use capabilities actually do the thing — acting across browsers, apps, and desktops for multi-step enterprise workflows. That’s worth a moment of consideration. Google stresses opt-in controls throughout: Connected Apps can be disconnected at any time, enterprise admins get policy levers via GeminiActOnWebSettings, and auto browse explicitly asks for confirmation before completing a purchase or posting on your behalf. Broader regulatory pressure is also building — Europe has moved to restrict Google from handling sensitive government, health, financial, and legal data, a context worth keeping in mind as AI agents gain deeper access to your browser. As Google put it in its official announcement, “We’re introducing major updates to Gemini in Chrome… that help you get the most out of the web.” Whether that framing reassures or raises an eyebrow probably depends on how much of your screen you’re comfortable sharing.
The shift here isn’t just a new button. It’s the browser evolving from a place you navigate into a system that navigates with you — and Chrome 149 is already live. Worth trying before deciding how you feel about it.
