Google isn’t giving up on the AI hype train. After integrating Gemini into Chrome, it wants to improve the AI chat interaction with Contextual tasks, which could allow the browser to do browsing for you. It’s a hidden experimental feature in Chrome Canary that Google hinted at a few months back.
Windows Latest force-enabled the Contextual tasks flag in the experimental section to test its capabilities. After that, a Contextual tasks option appeared in the More Tools section of the context menu. When I clicked on it, a sidebar appeared with the default Google homepage.
Browser researcher Leo also spotted these flags. Based on these flags and a couple of other references, it looks like Google is building a feature similar to “Actions in Edge.”
Gemini in Chrome for Windows 11 might automate long tasks. In our tests, Windows Latest observed that the sidebar doesn’t have much to show except the home page.
I clicked on the AI mode button, and it opened the Gemini page, which was weirdly cut off, making it impossible to type something. It looks a lot like the Copilot sidebar in Microsoft Edge, but in a less polished version. I wouldn’t be surprised if there’ll be an interface (sidebar?) to control Gemini inside Chrome and execute “actions.”
Its UI is all broken and doesn’t resize properly, but that makes sense given the feature is under internal testing.
However, do we have solid evidence that agentic capabilities for Chrome are in development?
Google hinted at agentic capabilities for Chrome
While the flags and our access to the AI sidebar make it quite obvious, Google’s blog post suggested that agentic browsing is coming to Chrome.
Google mentioned that agentic capabilities could solve problems like ordering groceries. It would reduce a 30-minute time waste into a three-click journey. But how? Since Gemini can already analyze open tabs data, it would suggest what to buy in a concise manner.
It could also retrieve past orders from your browser history or account data, but how good is that for privacy? Google will not tell you that.
Edge already has Copilot mode, which works similarly and can analyze tabs. I opened Crucial RAM buy product page, and it suggested the two options available, meaning it could read the pages and understand my intent.
Google is also working on multi-Gemini instances, which can help you use the AI assistant in each tab. Rather than asking Gemini to focus on one tab, you can run multiple AI queries in a single browser session without launching a new window.
