YouTube is moving to give users more direct control over what they see on the platform, announcing a new feature that lets people describe the kind of content they want recommended to them using plain, conversational language. The feature, called custom feeds, was unveiled at the company's Made On YouTube event on Wednesday and represents one of the more significant shifts in how the platform surfaces video content to its users.

Rather than relying entirely on YouTube's existing recommendation engine, users will now be able to type a description into a prompt box to define what a particular feed should contain. The input can be as specific or as broad as the user prefers. YouTube offered a few examples to illustrate the range of possibilities: someone might create a feed of video podcasts timed to fit a 30-minute commute, or they might request content suited to a particular mood, such as relaxing commentary videos for winding down at the end of the day.

Powered by Google's Gemini AI

The feature is built on top of Google's Gemini AI model, which processes the user's description and constructs a feed around it. Once created, the custom feed appears as its own tab pinned to the top of the YouTube home page, sitting alongside the platform's existing recommendation interface. Because Gemini handles the interpretation, users are not limited to short keyword-style prompts. They can write out detailed instructions covering what the feed should include, what it should leave out, which types of content to prioritize, and any other preferences they want the system to take into account.

The announcement positions YouTube within a broader trend across social media and content platforms, where companies have been racing to give users more agency over their algorithmic experience. Rather than treating recommendation systems as invisible black boxes, several platforms have begun exposing controls that allow people to describe and shape what they see.

A Trend Gaining Momentum Across Platforms

Bluesky, the decentralized social network, was among the first to make this concept widely visible, introducing customizable feeds as a key differentiator from competitors like X. The platform leaned heavily on user-built feeds as a way to demonstrate the flexibility its decentralized model could offer. Earlier this year, Bluesky took that further by rolling out an AI-powered tool called Attie, designed to make the feed-building process even more accessible to users who may not be technically inclined.

Other major platforms have followed suit. Meta's Threads, Instagram, and X have each launched their own variations of feed customization tools. More recently, Spotify joined the group with a comparable feature. Many of these implementations are also driven by AI, reflecting an industry-wide shift toward using large language models to interpret user intent and translate it into content curation logic.

Navigating a Library of Over 20 Billion Videos

YouTube's motivation for introducing the feature, according to the company, is rooted in the sheer scale of content available on the platform. Emily Moxley, Vice President of Product Management for Viewer AI at YouTube, addressed reporters ahead of the public announcement and emphasized just how large the platform's library has grown.

"One of the amazing things about being on YouTube is that the world is always at your fingertips. There's over 20 billion videos in the YouTube corpus, so you're always one search away from a treasure trove of topics."

The custom feeds feature, Moxley suggested, is intended to help users tap into that breadth more effectively. With so much content available, the standard recommendation algorithm can only do so much to surface material that aligns with a user's specific interests or needs at any given moment. By allowing users to define feeds themselves, YouTube is betting that people will discover more of what the platform has to offer.

A Complement to the Main Feed, Not a Replacement

It is worth noting that custom feeds are designed to work alongside YouTube's existing recommendation system, not to replace it. The platform's main feed, which aggregates recommendations based on a user's viewing history and engagement patterns, will remain in place. Custom feeds will appear as separate tabs, giving users the option to switch between their personalized algorithm-driven recommendations and the more deliberately constructed feeds they create themselves.

This approach allows YouTube to preserve the functionality that has driven much of its engagement over the years, while layering on additional tools for users who want more deliberate control over their experience. For casual viewers, the main feed will likely remain the default entry point. For those with more specific interests or use cases, the new tabs offer a way to go straight to curated content without having to search each time.

Rollout Timeline

YouTube confirmed that the ability to create multiple custom feeds will become available to users across both the web and mobile versions of the platform beginning next month. The company did not specify whether the rollout would be staged by region or user group, but indicated it would be a broad release covering both major access points for the service.

The feature marks another step in YouTube's ongoing effort to apply AI capabilities directly to the viewer experience, following a series of other AI-driven updates the platform has introduced over the past year. As competition for viewer attention continues to intensify across streaming and social video platforms, tools that give users a greater sense of control over their experience are increasingly becoming a standard expectation rather than a novelty.