YouTube will let users build custom feeds with Gemini

YouTube says it is introducing custom feeds that let users describe, in their own words, the videos they want to see. The prompt-based feature uses Google’s Gemini AI model to build a personalized feed around each request.

Users can ask for a feed focused on a particular interest or purpose. YouTube’s examples include video podcasts for a 30-minute train commute and relaxing commentary videos for unwinding. Prompts can also specify what the feed should feature, exclude, or prioritize.

Each custom feed will appear in its own tab pinned to the top of the YouTube home page. The feeds will supplement YouTube’s main recommendation feed rather than replace it, allowing users to switch between feeds based on their needs.

YouTube says the feature is intended to help people explore the vast amount of content on its platform. Emily Moxley, VP of Product Management for Viewer AI, said during a pre-brief that YouTube has over 20 billion videos in its corpus.

Support for creating multiple custom feeds will roll out on YouTube’s web and mobile apps starting next month, according to the company. The move follows similar feed-building tools from services including Bluesky, Meta’s Threads, Instagram, X, and Spotify.