Bluesky leans into AI with Attie, an app for building custom feeds

The team behind the decentralized social network Bluesky has unveiled Attie, a new standalone AI assistant designed to give users control over their content algorithms. Announced at the Atmosphere conference, the application allows individuals to build and curate custom social media feeds using natural language commands. The move signals a distinct strategy in the social media space, positioning AI as a tool for user empowerment rather than a mechanism for platform-controlled engagement optimization, putting algorithmic creation directly into the hands of its community.

Attie operates on Bluesky’s foundational AT Protocol and utilizes Anthropic’s Claude model to interpret user requests. The project is led by Chief Innovation Officer Jay Graber, who recently transitioned from the CEO role to focus on building new products. Conference attendees will serve as the initial beta testers. The announcement was coupled with news that the company secured $100 million in funding last year, providing a runway of over three years. This financial stability is intended to reassure developers and users of the protocol's long-term viability as the team explores monetization options like subscriptions or hosting, while explicitly ruling out any crypto integrations.

By launching a separate AI application, Bluesky is promoting its AT Protocol as a versatile foundation for a new class of user-centric software, not just a single social network. Interim CEO Toni Schneider drew parallels to the WordPress ecosystem, envisioning a future where numerous independent apps and services can coexist and interact on an open standard. This approach presents a clear alternative to the closed ecosystems of major technology platforms, aiming to foster a market of AI tools that serve individual needs rather than a platform’s internal metrics. In the long term, Bluesky plans for Attie to enable users to build entire social applications and tools for others.

Bluesky's launch of Attie is a strategic attempt to frame its AT Protocol as an open, user-programmable layer for social applications, directly challenging the closed, platform-centric AI models of its competitors by turning the algorithm into a feature users can build themselves.