YouTube lets users build custom feeds with Gemini

YouTube's custom feeds arrive as several other social platforms have already introduced tools that let users shape their own algorithms by describing what they want to see. Bluesky began popularizing this idea years ago as a way to differentiate its service from rivals like X, and this year it rolled out an AI tool, Attie, that makes feed-building even easier. Meta's Threads, Instagram, X, and more recently Spotify have also launched their own feed-building features, many of which are AI-powered.

YouTube's version uses Google's Gemini AI model to interpret lengthy, detailed prompts about what a feed should feature, exclude, or prioritize. The resulting feed is pinned to its own tab at the top of the home page, but does not replace the main feed of recommendations. Emily Moxley, VP of Product Management for Viewer AI, noted in a pre-brief that there are over 20 billion videos in the YouTube corpus, framing custom feeds as a way to help users explore that vast library.

The stakes for YouTube likely hinge on whether AI-built feeds feel accurate and useful enough that viewers actually use them alongside the default feed. If they do, it could shift some viewing away from the main algorithmic recommendations, which matters for creators whose reach depends on that system. Much will depend on how well Gemini translates casual descriptions into feeds that match what users had in mind.