YouTube announced custom feeds on Wednesday, allowing users to build their own AI-driven recommendation algorithm by typing text descriptions. Powered by Google's Gemini AI model, the feature lets users curate feeds for specific interests, moods, or commute lengths by detailing what content to prioritize, feature, or exclude. The resulting custom feeds will be pinned to the top of the YouTube home page in dedicated tabs, though they will not replace the main recommendations feed. Support for creating multiple custom feeds begins rolling out across web and mobile platforms next month.
The prompt-based feature follows similar algorithmic curation tools introduced by other social networks and platforms, including Bluesky, Meta's Threads and Instagram, X, and Spotify. Emily Moxley, VP of Product Management for Viewer AI, noted that the tool helps users explore YouTube's corpus of over 20 billion videos. Users can share lengthy, detailed descriptions to tailor the feeds to their exact preferences.
This feature matters because it gives YouTube users granular control over their viewing experiences through generative AI, joining a broader industry trend toward customizable algorithmic feeds.