AI
YouTube Launches Custom Feeds to Let Viewers Describe Their Own Recommendations
The video platform is now enabling users to create personalized feeds by typing natural language prompts that its AI system uses to curate content.
TechCrunch — AI
YouTube is rolling out a new capability that lets viewers craft individualized feeds by simply describing what they want to watch. The platform announced the custom feeds feature, which taps Gemini AI to interpret user requests and populate a dedicated tab on the YouTube home page with relevant video recommendations.
How It Works#
Users compose a text description of their desired feed—whether for a commute, a particular mood, or a specific topic—and submit it through a prompt box. The AI then constructs the feed based on these instructions. Because the system processes natural language, users can include detailed guidance about what to show, what to exclude, content themes, and presentation preferences. Once created, the feed appears as a pinned tab at the top of the homepage for quick access.
Examples YouTube offers include a podcast-focused feed for a 30-minute train ride or relaxing commentary videos for unwinding after work.
A Growing Trend#
YouTube's move mirrors recent initiatives across social and entertainment platforms. Bluesky championed this concept years ago and has since released Attie, an AI-powered feed builder. Threads, Instagram, X, and Spotify have all introduced similar features, many leveraging AI to streamline feed creation.
What This Means for Discovery#
YouTube hosts more than 20 billion videos, and the company sees custom feeds as a way to help viewers navigate this enormous catalog more efficiently. These new feeds complement rather than replace YouTube's main recommendation feed—users can switch between feeds depending on their needs when opening the app.
The feature will be available on both web and mobile starting next month.

