X just launched Custom Timelines — a major new feature that lets you pin hyper-personalized, topic-specific feeds directly to your Home tab. Instead of relying only on the general For You or Following tabs, you can now select from over 75 topics with more coming and pin a dedicated timeline for that niche.
Examples include things like art, finance, sports, tech, or whatever you’re deeply into. Each Custom Timeline is powered by Grok:
Grok understands the content of virtually every post on X. It combines that with X’s existing personalization algorithm. A feed that’s tailored just for you around that single topic — and it gets even better if you already engage with that subject a lot. This makes it easier to dive deep into specific interests without the mix of unrelated content that sometimes floods the main timeline.
Early access right now for Premium subscribers on iOS. Android rollout coming very soon. Announced today by X’s Head of Product, Nikita Bier, as one of the platform’s biggest changes in a while. This feels like a smart evolution: more control over what you see, less noise, and deeper engagement in the niches that matter to people. Grok’s role in classifying content at scale is what makes the personalization actually work at this level.
Grok’s content classification on X is the core AI capability that powers features like Custom Timelines, the personalized For You feed, and ranking in the Following tab. Grok doesn’t just scan for keywords or hashtags. It semantically understands the meaning, context, nuance, and intent of virtually every post on X — including text, images, videos, threads, replies, and memes.
Register for Tekedia Mini-MBA edition 20 (June 8 – Sept 5, 2026).
Register for Tekedia AI in Business Masterclass.
Join Tekedia Capital Syndicate and co-invest in great global startups.
Register for Tekedia AI Lab.
For each post, Grok performs real-time analysis to determine: Main topic(s) — e.g., is this about Formula 1, AI ethics, Sub-niches and granularity — It can distinguish deep sub-topics. How substantive, original, or engaging the content is. Sarcasm, humor, misinformation risks, or cross-topic connections. This understanding is then combined with X’s traditional personalization engine (your past likes, replies, follows, dwell time, etc.) to decide what shows up where.
When you pin a topic; one of 75+ available, like Art, Finance, Tech, Sports, Memes, etc: Grok filters the entire firehose of X posts. It classifies which posts belong to your chosen topic using its deep semantic model not rigid rules. It personalizes the feed further: posts you’re more likely to engage with based on your history in that niche rank higher.
The result is a clean, dedicated timeline focused only on that topic — and it gets smarter and more precise the more you interact with that subject. This is why the official description says: It’s powered by Grok’s understanding of every post with the algorithm’s personalization—meaning every timeline is made just for you. And it works even better when it’s a topic you already engage with.
Grok helps recommend posts beyond who you follow by classifying content and matching it to your inferred interests. Even posts from accounts you follow are now ranked not purely chronological using Grok’s predictions of what you’ll find engaging. Topic snooze, content moderation signals, spam/low-quality filtering, and future prompt-based feed adjustments all lean on this same classification layer.
Traditional social media algorithms rely heavily on engagement metrics likes, retweets, views + basic topic tagging. Grok shifts toward true understanding: It handles nuance; context, sarcasm, evolving slang. It scales to hundreds of millions of posts daily. It reduces noise in niche feeds because it’s not just matching keywords — it’s comprehending meaning.
The system is still evolving. X has been moving toward a purely AI algorithm, with Grok playing a central and growing role in ranking and filtering. In short Grok acts as X’s intelligent content layer — reading and categorizing posts at massive scale so the platform can serve you hyper-relevant, topic-specific experiences instead of one generic algorithmic soup.