What’s happened? Google is testing a Chrome feature called “Contextual tasks” in its Canary build. When enabled via hidden flags, it adds a More Tools entry that opens an unfinished side panel tied to Gemini. The panel currently defaults to the Google homepage or a cut-off Gemini view and does not resize correctly.

• Spotted by Windows Latest and testers on X, force-enabling the feature shows us a sneak peek of what Google has in store in the future.
• Google has hinted at its plans for Chrome and AI, with contextual suggestions in the US first, with broader agentic flows under active development.
• The design echoes Edge’s Copilot rail, hinting at deeper in-browser assistance.

This is important because: This looks like groundwork for a browser that can actually perform steps for you. Hosting Gemini inside Chrome creates a path from a typed request to real clicks and entries across the web.

• It signals native in-browser steps, letting the prompt perform the navigation and complete forms.
• A built-in command panel cuts the copy-paste middleman, a common failure point for web assistants.
• Google has said tasks like grocery orders could shrink from half an hour to a few clicks, and this is the plumbing that could make that happen.

Why should I care? Usefulness comes with strings attached. To act on your behalf, an assistant needs permission to read pages, juggle tabs, and interact with data tied to your account. You get convenience, but you also give up potentially sensitive info.

• Automating tasks means letting AI read what is on screen and sometimes past context.
• Permissions can sprawl, from page content and form fields to cookies or session state.
• If bots handle more activity, sites may see fewer human visits and less direct engagement.
• Users still want an obvious stop button, plus granular scopes they can revoke quickly.

Okay, so what’s next? Microsoft is further along in Edge and Copilot, while Chrome is laying the track in Canary. Expect conservative rollouts and small opt-in steps that expand as controls mature. We’ve tested AI browsers like Dia and Comet to give you a glimpse of the AI browsing experience.

• From our time with Dia, an AI-first browser can feel effortless on repetitive tasks, but it raises bigger questions about data handling and reliability that Chrome and Edge must answer.
• In our Comet test, comparisons and summaries were faster, but errors sometimes erased the gains.