Bluesky Launches Attie: AI-Powered Custom Feed Builder for the Open Social Web

Bluesky has built its reputation on a specific promise: the infrastructure of social media should be open, portable, and legible to users. Algorithms that determine what you see should not be black boxes operated by a corporation with undisclosed commercial interests. The platform's atproto protocol is designed to make feed curation a user-controlled activity rather than a platform-imposed one.

Attie is the company's attempt to make that philosophy accessible to non-technical users. Using natural language, a Bluesky user can describe the kind of content they want to see — by topic, tone, source type, or behavioral characteristic — and Attie translates that description into a working custom feed powered by atproto's open feed generator infrastructure.

Why Custom Feed Builders Matter for Open Social

The ability to create custom feeds has existed on Bluesky since the platform's early days, but it has always been a technical capability requiring programming knowledge. Feed generators — the components that define what content populates a given feed — can be built by anyone, but building one required understanding atproto's data schemas, hosting a service endpoint, and writing query logic.

Attie removes that barrier. A user who wants a feed of only original long-form writing from journalists in their professional network, excluding reposts and articles older than 48 hours, can describe that in plain English. Attie handles the technical translation.

The significance of this extends beyond convenience. One of the structural advantages claimed for open social protocols is that users can escape algorithmic intermediation — they can define their own information environment rather than having it defined for them. But that advantage has largely been theoretical for users without technical skills. Attie turns it practical.

AI as Infrastructure, Not Algorithm

There is an important distinction in how Attie uses AI compared to how recommendation algorithms work on closed platforms. A recommendation algorithm on TikTok, Instagram, or X is optimizing for platform-defined objectives — typically engagement metrics that correlate with advertising revenue. The algorithm is an expression of the platform's commercial interests as much as the user's preferences.

Attie uses AI differently: as an interpreter of user intent, not an optimizer for platform objectives. The AI's job is to accurately translate what the user describes into a feed configuration — not to decide what the user should want. The commercial interests that shape how AI is deployed here are the user's interests, not Bluesky's revenue model.

This is a more defensible use of AI in social infrastructure, and it is worth noting as a design pattern that other open-protocol applications may adopt.

Competitive Context: AI Features on Social Platforms

Every major social platform is integrating AI features, typically in ways that increase platform control over the user experience — AI-generated summaries, AI-recommended content, AI-moderated feeds. These features are presented as user benefits but structurally concentrate power with the platform operator.

Bluesky's Attie inverts that dynamic. The AI serves user agency rather than platform control. Whether that positioning proves commercially viable — whether enough users care about algorithmic transparency to choose Bluesky over more feature-rich closed platforms — is the central question the product must answer.

Bluesky currently has approximately 30 million registered users, a fraction of X's or Instagram's scale. Attie alone will not close that gap. But it represents a coherent vision of what AI-native social infrastructure can look like when built on open principles rather than closed commercial ones.