Editorial illustration for YouTube's New AI Feature Lets You Customize Your Feed With Detailed Descriptions
YouTube Lets Users Build Custom Feeds With AI
YouTube's New AI Feature Lets You Customize Your Feed With Detailed Descriptions
YouTube announced on Wednesday that it's giving users a way to talk directly to the algorithm instead of just feeding it clicks and watch time. The feature, called custom feeds, lets people type out a description of the videos they want to see and have YouTube build a dedicated feed around it. Want something for a 30-minute train commute?
Type that in. Want relaxing commentary videos to unwind with after work? That works too.
The system runs on Google's Gemini AI model, which takes the written prompt and translates it into a curated feed that then gets pinned to its own tab at the top of the YouTube home page. Because it's AI-driven, users aren't limited to a few keywords. They can get specific about what to include, what to leave out, and what to prioritize, turning a vague preference into something closer to a set of instructions.
YouTube isn't alone in trying this. Other platforms have been rolling out similar tools that hand users more direct control over what shows up in their feeds, rather than leaving it entirely to opaque recommendation systems.
YouTube says it’s going to let you build your own algorithm — well, kind of. The company on Wednesday announced a new feature called custom feeds that allows you to describe, in your own words, the kinds of videos you’d like YouTube to recommend to you.
Why this matters
Custom feeds is YouTube's version of a trend that's already played out on X with Grok-built timelines and on Instagram with interest-based feed tweaks. The pitch is control: describe what you want, skip what you don't, get a feed that isn't just whatever the black-box algorithm decided you'd click. For founders building on top of recommendation systems, this is a signal worth watching.
Natural-language steering is becoming the expected interface for personalization, not a novelty feature. If YouTube can make prompt-based curation work at its scale, users on smaller platforms will start asking why their feeds still require the pull down and hope method.
The catch is trust. YouTube's core algorithm still decides what qualifies as a match for your description, and the company hasn't said how much visibility users get into why a video showed up or got excluded. A custom feed built on an opaque ranking system is still an opaque ranking system, just with a nicer front door. Worth tracking whether YouTube publishes any detail on how these prompts actually get translated into ranking signals, because that's where the real product is.
Related Reading
Business & Startups Nordic pilot adds Gemini for Education, NotebookLM to boost AI literacy
Business & Startups Kling launches Video O1, all-in-one model with MVL bridge using transformer
Business & Startups DeepSeek Seeks More Capital Weeks After USD 7B Funding Round
From Archives Basecamp's AI Training Dataset to Reach One Quadrillion Tokens
From Archives Alibaba Slashes AI Audio Prices 95%, Launches Qwen 3.1 Models
Common Questions Answered
How does YouTube's custom feeds feature work with natural language descriptions?
YouTube's custom feeds allow users to type detailed descriptions of the types of videos they want to see, and the system uses Google's Gemini AI model to interpret these written descriptions and build a dedicated feed around them. Instead of relying solely on clicks and watch time, the algorithm now processes user-provided text to understand and deliver personalized content recommendations.
What are some practical use cases for YouTube's custom feeds feature?
Users can create custom feeds for specific situations, such as requesting videos suitable for a 30-minute train commute or relaxing commentary videos to unwind with after work. The feature enables people to describe their exact preferences in their own words rather than waiting for the algorithm to learn their viewing patterns through implicit signals.
How does YouTube's custom feeds compare to similar AI features on other social platforms?
YouTube's custom feeds is part of a broader trend where social media platforms are adopting natural-language steering for personalization, similar to X's Grok-built timelines and Instagram's interest-based feed tweaks. This shift represents a move toward giving users more explicit control over their feeds rather than relying on black-box algorithms to determine what content they see.
Why is natural-language steering becoming important for recommendation systems?
Natural-language steering is becoming the expected interface for personalization because it gives users direct control over their content feeds by allowing them to describe exactly what they want to see. This approach addresses the limitations of traditional algorithmic recommendations that rely on implicit signals like clicks and watch time, providing a more transparent and user-directed experience.
Further Reading
- YouTube is working on a feature that will fix the messy home feed - TechCrunch
- YouTube will let you ask AI to make a custom video feed - The Verge
- YouTube's AI Is Ready to Customize Your Scrolling - CNET
- YouTube now lets you create a 'custom feed' about anything you want - Android Authority
- YouTube now lets you ask AI to build a video feed for you - Mashable
