Google Discover is getting an AI chatbot-tuned feed

Google is rolling out a conversational AI feature that lets users shape their Discover feed through natural language. By accessing a chatbot interface in the app's menu, users can describe their content interests, and the system will remember those preferences to personalize future feed results automatically.

Google is introducing a natural language customization tool for its Discover feed, set to arrive in the Google app within the next few days. Instead of manually toggling topics, users will be able to simply describe what kind of content they want to see, and an AI-powered chatbot will interpret those preferences and adjust the feed accordingly.

The feature is tucked inside the three-dot menu on the Discover feed and opens a conversational interface where users can refine their choices iteratively if the first attempt misses the mark. Crucially, the system is designed to retain those stated preferences over time, meaning users won't need to repeat themselves on every visit. This represents a meaningful shift from passive algorithmic curation toward active, user-directed personalization.

Google is upgrading its Discover feed — the personalized content stream surfaced on the Google app's home screen — with a conversational AI customization tool expected to launch within days. Rather than relying solely on implicit signals like browsing history and click patterns, users will now be able to explicitly tell the system what they want through plain language.

The feature lives inside the three-dot overflow menu on the Discover feed and opens a chatbot-style panel. Users type out their preferences in natural language, the AI acknowledges them and describes what content it intends to prioritize, and users can push back or add nuance if the interpretation isn't quite right. That back-and-forth dynamic mirrors the kind of iterative prompting people are growing accustomed to with tools like ChatGPT or Gemini.

One of the more practical elements is memory. The system is built to retain a user's stated preferences across sessions, which removes the friction of re-explaining interests every time. This positions Discover less as a static recommendation engine and more as a living, dialogue-driven service that evolves alongside the user's stated needs.

Why it matters: This move signals how deeply conversational AI is being woven into everyday Google products, far beyond standalone chatbots. Search, discovery, and content curation are all quietly being restructured around the idea that users can — and increasingly prefer to — express their needs in natural language rather than through menus and toggles. For Google, it's also a way to differentiate Discover from social and news aggregator competitors by making personalization feel more transparent and controllable.

That transparency angle could prove significant. One frequent criticism of algorithmic feeds is that users feel trapped inside invisible preference bubbles with no clear way out. Giving people a direct conversational lever to reshape what they see may help Google address that frustration, even as questions remain about how faithfully the AI actually honors those stated preferences over time.