YouTube Adds Gemini-Powered Custom Feeds and Shopping Search

YouTube announced custom feeds, a feature that lets users type a natural-language description of the videos they want recommended, with Google’s Gemini model building the feed and pinning it in a dedicated tab atop the YouTube home page. [1] Users can provide lengthy instructions covering what the feed should feature, exclude, or prioritize, while suggested uses include podcasts for a 30-minute commute and relaxing commentary videos for unwinding. [1]

The addition follows custom-feed tools from other social networks, including Bluesky’s AI-powered Attie, as well as offerings from Threads, Instagram, and X. [1]

At the Made On YouTube event, the company said Ask YouTube would become a shopping companion, extending a conversational search feature that accepts detailed natural-language queries and compiles results from Shorts and long-form videos. [2] Product searches may produce organized recommendations and comparison tables based on user preferences, while viewers will also be able to ask follow-up questions directly from a video’s watch page. [2]

YouTube said more than 140 million people used Ask YouTube to explore videos more deeply during June, representing growth of more than 500% since the previous December. [2] The feature uses a custom Gemini model that maps complex viewer questions to the exact second of a video containing the answer, according to YouTube executive Emily Moxley. [2]

YouTube is applying prompt-based control at two different stages of discovery: custom feeds shape what appears on the homepage, while Ask YouTube interprets questions within search, video viewing, and shopping. The feed feature places YouTube alongside Bluesky, Threads, Instagram, and X, but the shopping expansion also connects natural-language discovery with organized product comparisons and creator videos. What the sources don't address: how accurately custom feeds honor exclusions or how YouTube ranks and discloses shopping recommendations.