YouTube announced a new feature, custom feeds, that lets users shape their own recommendations by writing a plain-language description of what they want to watch. The prompt-based feature, announced Wednesday at the Made on YouTube event, uses Google’s Gemini AI model to build a feed around the request. Once created, the feed appears as its own tab pinned to the top of the user’s YouTube home page.

To create a custom feed, a user types a description into a prompt box, such as a feed dedicated to a particular interest or one with a specific purpose. YouTube’s examples include a feed of video podcasts to watch during a 30-minute train commute, or a request for a mood-driven feed like “relaxing commentary videos to unwind with.” Because the feature is AI-powered, the request can be detailed: users can specify what the feed should feature or exclude, and what content to prioritize, in lengthy terms.

The move follows similar tools from other platforms. Decentralized social network Bluesky popularized the idea of letting users build their own algorithms and this year rolled out an AI tool, Attie, that simplifies feed-building. Meta’s Threads, Instagram, X, and more recently Spotify have also launched feed-building features, many of them AI-powered.

YouTube frames custom feeds as a way to help users explore the scale of its catalog. “One of the amazing things about being on YouTube is that the world is always at your fingertips,” Emily Moxley, VP of Product Management for Viewer AI, said in a pre-brief before the event. “There’s over 20 billion videos in the YouTube corpus, so you’re always one search away from a treasure trove of topics.”

The new feeds will not replace YouTube’s main recommendation feed, which remains in place. Instead, they give users an additional way to jump directly into a feed tailored to a specific purpose when opening the YouTube app. Support for creating multiple custom feeds will roll out across YouTube on web and mobile starting next month.