According to Windows Report | Error-free Tech Life, Google is testing a radically new version of its AI Mode in the Chrome Canary browser. The feature no longer redirects users to a Google Search page but instead opens a built-in interface at chrome://contextual-tasks. This native page can answer questions, read the content of open browser tabs, process uploaded files like PDFs, and generate images. The tool maintains a continuous conversation, remembering context across different tasks, such as generating a pizza image and then answering questions about its ingredients. The version is still unfinished, with placeholder labels like [i18n] Ask Google…, and Google has not announced a release timeline for stable Chrome users.
The big shift from web to native
Here’s the thing: this is a fundamental architectural change. Right now, most browser “AI features” are just fancy webpages. You click a button, and you’re whisked away to a chat interface that lives on a server somewhere. But this new Chrome Canary build? It’s baking the AI directly into the browser’s guts. That chrome://contextual-tasks address is the tell. It’s a privileged, internal page, not a remote website.
So why does that matter? Well, for one, it can potentially be faster and work offline (eventually). More importantly, it has much deeper, more secure access to your local browser context—your open tabs, your downloaded files. Asking it to “summarize this PDF” and having it just work, without you ever uploading that file to a cloud server, is a different privacy and usability proposition. It feels less like using a web service and more like using a feature of your computer.
The all-in-one context machine
The most impressive demo from the report is the seamless context switching. The AI can read a webpage, then you drop in a PDF, and it knows to focus on the file for the next question. Then you can ask it to create an image and *ask questions about the image it just made*. That’s wild. It’s not treating each query as an isolated event. It’s maintaining a session that understands the mix of text, files, and images you’ve thrown at it.
This is where Google hopes to have an edge over standalone chatbots. Your browser is already the central hub of your digital work—it has all your tabs, your downloads, your history. Turning it into an AI that understands that specific, personal context is a powerful idea. Basically, they’re trying to make the browser itself intelligent, not just putting an intelligent website inside the browser.
Rough edges and big questions
Now, let’s be clear: this is in Canary. That’s the absolute bleeding-edge, often-broken version of Chrome for developers. The placeholder text proves it’s early days. But the core functionality is already working, which tells us Google is serious about this direction.
The big questions are about cost and capability. Running this level of AI locally is computationally expensive. Will it require a specific hardware tier? Or will most of the heavy lifting still happen in Google’s data centers, just with a nicer local interface? And what models will it use? The ability to analyze a generated image suggests vision capabilities are baked in from the start.
I think we’re seeing the blueprint for Chrome’s future. They’ve already been testing AI on the New Tab Page and the address bar. This contextual tasks page looks like the unifying hub for all of it. It’s not just an AI feature anymore. It’s becoming the browser’s new core interface.
