YouTube Launches Gemini AI Custom Feeds to Build Your Own Algorithm

Executive Overview

YouTube has officially unveiled Custom Feeds, a new feature that empowers users to build their own recommendation algorithm using natural language descriptions powered by Google's Gemini AI. Announced at the Made On YouTube event, the feature will roll out on web and mobile starting October 2026, free of charge. It arrives amid escalating competition among social platforms—Bluesky, Threads, Instagram, X, and Spotify—to give users control over their content algorithms. YouTube's entry is distinguished by its integration of Gemini AI and its vast library of over 20 billion videos.

Deep-Dive Features & Architecture

The Custom Feeds feature operates through a simple interface: users type a text description into a prompt box within YouTube, specifying what they want the feed to contain. For example, YouTube suggests creating a feed for podcast video streams to watch during a 30-minute train ride, or requesting videos for a specific mood such as "relaxing commentary videos for unwinding." The system uses Google's Gemini AI model to analyze the description and build the feed, which is then pinned in a dedicated tab at the top of the homepage. The feature supports long-form, detailed prompts that allow users to specify what to include or exclude and to prioritize certain content.

Emily Moxley, YouTube's VP of Product Management for Viewer Intelligence, said during a press briefing ahead of the Made On YouTube event: "The amazing thing about YouTube is that the world is always at your fingertips. There are over 20 billion videos in YouTube's library, so you're always one search away from a treasure trove of topics." She explained that the new feeds will not replace the main feed but will allow users to quickly switch between the feed best suited to their needs when opening the app. The company announced that support for creating multiple custom feeds will begin rolling out on web and mobile starting next month.

Benchmark & Competitive Performance

YouTube's feature arrives in the context of escalating competition over "user algorithms" among social media platforms. Decentralized network Bluesky was among the first to promote the idea of custom feeds as a way to differentiate from X, and this year launched an AI-powered tool called Attie to simplify feed building. Threads, Instagram, X, and Spotify have also launched similar feed-building features, most AI-powered. What sets YouTube apart in this competition is its reliance on Google's Gemini AI model and its library exceeding 20 billion videos, making the biggest challenge not building the algorithm but sorting and accurately personalizing this massive volume of content.

Industry Impact & Enterprise Adoption

For developers and users, the introduction of Custom Feeds signals a broader shift toward user-controlled algorithms, potentially reducing reliance on opaque recommendation systems. For enterprises and content creators, this opens new avenues for targeted content delivery and audience engagement, as users can explicitly request niche topics or formats. The integration of Gemini AI also highlights Google's strategy to embed its AI models across its product ecosystem, reinforcing its competitive position against rivals like OpenAI and Meta. As platforms race to offer algorithmic transparency and control, YouTube's move could set a new standard for personalized content discovery at scale.

Conclusion

YouTube's Custom Feeds, powered by Gemini AI, represents a significant step in giving users direct control over their content algorithms. With a free rollout on web and mobile starting October 2026, and leveraging a library of over 20 billion videos, YouTube aims to differentiate itself in a crowded social media landscape. While specific benchmarks and security details remain undisclosed, the feature's natural language interface and long-form prompt support position it as a powerful tool for personalized content discovery. As competition intensifies, YouTube's integration of Gemini AI could redefine how users interact with algorithmic recommendations.

Media Source: TechCrunch AI | Fact Verification & Analysis: AI Tools Oasis