(Credit: Google/PCMag)
Google isn't giving up any ground to Apple or Microsoft in the race to add AI features to their respective platforms. Alongside a fresh class of Chromebooks with neural processing units (NPUs), Google introduced some intriguing new AI features for ChromeOS. Some work on every Chromebook, some only on Chromebook Plus machines, and some only on the new NPU-equipped models.
One new feature stood out to me amid all the announcements: Text Capture. With this Chromebook Plus tool, you can select any part of your screen with text, whether it's part of an image or by itself. Text Capture then proposes related actions and carries out the one you choose. It can, for example, create a calendar event with all the details from a selected image. It's similar to what Apple's Visual Intelligence promises for future versions of iOS and what Click to Do already does on Copilot+ PCs.
Text Capture is indicative of Google's strategy to make AI features as frictionless as possible. Google's vice president of product management for ChromeOS, John Maletis, said it’s about “bringing AI to where the user is.” That means offering AI tools, no matter what you're doing on your computer, without requiring you to open a separate application. Like Apple and Microsoft, Google is emphasizing on-device AI processing but tapping into the cloud for some of the more computationally intensive features.
I saw the Text Capture tool (and others) in action at Google’s SoHo offices in New York and came away impressed, especially since even the most advanced Chromebooks cost less than a competent Copilot+ PC, iPhone, or MacBook. Below, you can read my initial impressions of Text Capture and the other new ChromeOS AI features based on a live demo. I look forward to testing everything for myself soon.
How to Use Text Capture in ChromeOS
Despite not requiring one of the new Chromebooks with an NPU, Text Capture uses an on-device AI model. It works directly with any Chromebook app, meaning you don't have to stop whatever you are doing to launch it.
(Credit: Google/PCMag)To get started, long-press the Launcher key (the button in the lower-left corner of the desktop with the G) or the space bar key. An overlay message appears and instructs you to select anywhere on the screen by dragging your finger or dragging with a touchpad or mouse. During my demo, the Google rep selected an area of a grocery receipt with purchases and prices.
(Credit: Google/PCMag)Once you select something, you should see a panel that shows an AI overview about it. Otherwise, you can run a search on the image or text in question. That part works like the Google Lens AI feature I mention later. You get a colorful border animation around the panel, and a set of relevant action buttons appears. For the receipt, the options were Create Google Sheet, Copy Text, and Copy With Formatting. That first option seems the most useful, since a spreadsheet gives you tons of options for working with numerical data.
(Credit: Google/PCMag)After choosing the Google Sheets option, a brief “Creating” dialog appeared in the lower right side of the screen. Finally, Google Sheets opened with the selected text from the image all in neat rows and columns. Just note that you might need to clean up any text from an image that doesn’t fit a pattern, such as the “2 @ $1.25 ea” line below.
(Credit: Google/PCMag)You can also create a calendar entry from an image you select. Text Capture can tell if a selection contains event information, in which case you get a button that proposes adding a calendar appointment. (Below, you can see the colorful animation around the selection box.)
(Credit: Google/PCMag)Again, relevant buttons appear below your selection. After the Google rep chose Add to Calendar, a fully populated event entry appeared in Google Calendar.
(Credit: Google/PCMag)Text Capture vs. Similar Apple and Microsoft Features
Text Capture somewhat overlaps with the Visual Intelligence features Apple announced at WWDC, but there are some important differences. First, the Apple tool is available only on iPhones running iOS 26, meaning iPad and Mac users are out of luck. Second, with Apple, you have to take a photo or screenshot before you can do anything, whereas Google’s Text Capture lets you select anything that's on the screen. In its favor, Apple’s Visual Intelligence understands objects in images in addition to text.
Microsoft’s Click to Do feature also lets you select objects on your Windows 11 desktop. It then suggests relevant actions. Click to Do doesn’t take you out of your current activity either, requiring you to just press the Windows key and click on the object of interest. Click to Do can summarize or rewrite text with AI and open or edit images, but it lacks Text Selection’s ability to create a calendar event or spreadsheet based on a selection.
Even More New ChromeOS Features
Google also announced the Smart Grouping feature for the new range of Chromebooks with NPUs. This uses on-device AI to identify apps, tabs, and windows that are part of a related overall task and groups them together on a new virtual desktop.
The ability to generate AI images via the Quick Insert key is another cool exclusive for Chromebooks with an NPU. This works directly within an app, such as Google Docs. You have to give up the Caps Lock key to use this feature, though you can still get that functionality in other ways. Here's what the Quick Insert interface looks like in use:
(Credit: Google/PCMag)All ChromeOS users will get the new Select to Search With Lens feature: When you select an object in any application, a Google Lens panel with information about it pops up. This tool is also available from the screenshot utility.
The "Help me read" feature now includes a Simplify option that uses AI to condense or clarify complex text. Finally, Google's NotebookLM AI research tool will be preinstalled on Chromebook Plus laptops.
Can Google Deliver on Its Goal of Practical AI?
All the AI tools appearing on desktop OSes, whether from Apple, Google, or Microsoft, have the potential to make everyday computing easier. Google’s latest features seem competitive and genuinely useful so far, but I plan to test them more to see just how much they improve my efficiency.