Bluesky launches Attie — an AI assistant for building custom content feeds

What happened

On March 28, 2026, Bluesky announced Attie at the company’s Atmosphere conference. Attie is a standalone AI application built on the AT Protocol and powered by Anthropic’s Claude. It allows users to describe, in natural language, the type of content they want to see — and then constructs a personalized feed based on that description, without requiring any technical configuration.

Bluesky’s existing feed system allows users to subscribe to custom algorithmic feeds, but building those feeds previously required familiarity with AT Protocol development. Attie removes that barrier: a user can write a description like “original reporting about AI policy from journalists I don’t already follow” and Attie generates the corresponding feed.

Why it matters for writers and publishers

The relevance for writers and publishers is in distribution. Bluesky has been growing as an alternative to platform-controlled social networks, and Attie introduces a new mechanism through which readers discover content there. Because feeds are user-constructed from natural language descriptions rather than served by a single platform algorithm, writers whose work matches a reader’s stated interest have a direct path to reach that audience — without depending on algorithmic promotion decisions made by the platform.

For writers building an audience on Bluesky, understanding how Attie constructs feeds is practically useful: it means the signals that matter for discoverability are closer to descriptive relevance and topical specificity than to engagement metrics or posting frequency. This shifts some of the logic of content strategy on Bluesky compared to how attention-maximizing platform algorithms typically work.

The use of Anthropic’s Claude as the underlying model also makes Attie relevant for writers who track how AI systems interpret and surface written content — it is one of the first public examples of Claude being used specifically for content discovery rather than generation.