Personalising Video Recommendations with Natural Language
YouTube has officially announced a new AI powered feature known as Custom Feeds, designed to give viewers greater direct control over their video recommendations. Built upon Google's flagship Gemini artificial intelligence model, the system allows users to generate tailored video streams simply by typing natural language prompts.
Rather than relying purely on automated, implicit watch history signals, viewers can describe precisely what kind of content they want to see, prioritize, or exclude. For instance, users can request specific genres or contextual activities, such as requesting video podcasts tailored to a thirty minute commute or guided meditation sessions under ten minutes to unwind after work.
Seamless Integration Across Web and Mobile
The feature generates dedicated recommendation tabs that sit pinned atop the YouTube home page, running alongside the platform's traditional recommendation feed rather than completely replacing it. Viewers can seamlessly jump into these custom-curated tabs whenever they launch the application on web, mobile, or smart TV interfaces.
Users will be able to create, save, and edit multiple custom feeds across their devices starting next month. By offering dynamic text editing directly within the feed header, viewers can instantly adjust, refresh, or fine tune their prompt criteria as their viewing goals evolve throughout the day.
