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
In YouTube's latest update, viewers can tweak what the algorithm shows on their front page and curate it to a specific vibe.
Context & Ripple Effects
YouTube had already tested prompt-built, continuously refreshed home feeds for eligible US signed-in viewers in May; that earlier custom-feed rollout established the interaction model that this feature makes more durable through saved feeds. The move extends a long shift away from a single universal homepage algorithm, from YouTube’s socially oriented 2011 homepage redesign to its planned creator-channel “For You” sections in 2023.
The significance is distributional: viewers are being given a natural-language layer for stating what they want to watch, while creators’ reach on the homepage becomes more contingent on the intents those prompts express.
First-order effects
- US users can create and retain multiple tailored homepage feeds rather than relying solely on YouTube’s default recommendations.
- Creators’ videos may reach viewers through explicitly prompted feed themes, adding a user-directed route to homepage discovery.
Second-order effects
- Creators using YouTube’s newly announced A/B-testing tools have a stronger incentive to test packaging and formats against distinct audience intents, tightening the creator-specific recommendation and optimization loop.
- YouTube’s default recommender must coexist with saved, prompt-defined feeds, making homepage distribution less dependent on one generic ranking surface.
Third-order effects
- If prompt-defined feeds gain sustained use, video discovery shifts toward a hybrid model: platform recommendation supplies the inventory, while users increasingly set the context in which it is ranked.
- That change would make algorithmic eligibility more legible to creators and viewers without removing YouTube’s control over which videos are available to recommend.
The trend: Video platforms are adding conversational controls to algorithmic discovery, turning the homepage from a single personalized feed into a set of user-defined viewing contexts.