At its Made on YouTube event, YouTube unveils Custom Feeds, an LLM-powered feature that lets users generate and save tailored homepage video feeds, for US users

You can now use plain English to curate your YouTube homepage for specific moods or interests.

YouTube has introduced a new AI-powered feature called Custom Feeds, allowing US users to generate and save personalized video feeds. Unveiled at the Made on YouTube event, this tool enables users to type natural language requests, such as desired moods or specific video categories, and have YouTube's large language models create a tailored homepage. Users can access Custom Feeds via a button below the search bar, and the feature does not replace their standard YouTube homepage. This development follows successful testing of the feature earlier this year and expands user control over algorithmic recommendations beyond simple likes or blocks. The company, aiming to give users more agency in their content consumption, plans to roll out Custom Feeds to all US users on mobile, desktop, and TV soon.

YouTube's Custom Feeds feature marks a significant step in AI-driven user experience personalization, moving beyond basic engagement signals to incorporate nuanced user intent. By leveraging LLMs, YouTube offers a low-friction method for users to express complex preferences, directly influencing algorithmic output. This approach contrasts with earlier, more intimidating attempts to give users direct control over core recommendation systems, suggesting a market trend towards more accessible AI tools for personalization.

This rollout places YouTube alongside competitors like Meta and Spotify, who have also introduced generative AI features to allow users to customize their content feeds. The success of Custom Feeds could influence how other platforms balance algorithmic curation with user autonomy, potentially leading to a broader adoption of natural language interfaces for managing digital experiences across various media services.