Bluesky has added a feature called Quests to its AI assistant Attie, allowing users to ask natural-language questions about information and news propagating across Bluesky and any application built on the AT Protocol. The feature was announced the week of July 24, 2026 and is currently in beta, with users admitted from a waitlist. TechCrunch

Attie launched in March 2026 as a tool for building custom social feeds without programming. The addition of Quests shifts its role from feed construction toward open social research. A user can pose a question to Attie and receive answers synthesized from activity across the federated AT Protocol network, not just Bluesky's own app. That scope matters because the AT Protocol is an open standard; any third-party client or service that integrates with it contributes to the data surface Attie can query. The 45.6 million registered accounts on Bluesky as of July 2026 represent one node in that ecosystem, though by far the largest.

The leadership context is relevant. Jay Graber, who had served as CEO since Bluesky's inception, stepped down in March 2026 and moved into a chief innovation officer role. Toni Schneider replaced her as CEO. The Attie launch coincided with that transition, and Quests arrives roughly four months later under the new structure. Bluesky had closed a $100 million Series B round in 2025, giving the company runway to build out product surfaces like Attie beyond its core social client.

Attie is free at the time of the Quests announcement, though Bluesky has considered charging for the service. A paid tier for AI-mediated social research would fit a familiar pattern: offer a capability at no cost during a beta phase to establish usage patterns and gather feedback, then introduce monetization once the feature set stabilizes. The company is also redesigning Attie with new accessibility features and a refreshed logo, suggesting the product is moving from early experiment toward something intended for broader, sustained use.

The technical distinction worth noting is the AT Protocol's openness. Attie's Quests queries are not confined to Bluesky's firehose in isolation. Any app that speaks the protocol, hosts data on a Personal Data Server, and participates in the federation is part of the queryable graph. This means a researcher asking Attie about a trending topic could surface posts, interactions, and metadata from clients that are not Bluesky-branded at all. The protocol-level architecture is what makes Quests functionally different from a search box inside a single platform.

What remains unclear from the announcement is how Attie handles source attribution, provenance, and the reliability of synthesized answers. Social platforms are noisy environments where misinformation, satire, and coordinated campaigns coexist with genuine reporting and analysis. An AI assistant that summarizes "information spreading across" a federated network inherits all of those signal-quality challenges. The beta rollout through a waitlist gives Bluesky room to iterate on those issues before broader access.

In this author's view, the more interesting long-term question is whether Attie becomes a research tool that third-party developers and organizations build on top of, or whether it remains a Bluesky-native feature that happens to query open protocol data. The AT Protocol was designed to decentralize social media infrastructure; an AI layer that makes that infrastructure queryable in natural language could accelerate the ecosystem effect, drawing more apps and users toward the protocol. That depends on whether Bluesky exposes Attie's capabilities via API, what rate limits or access tiers apply, and whether the monetization strategy targets end users, developers, or both. None of those details are public yet.

The redesign and accessibility push suggest Bluesky is treating Attie as a product line rather than an experiment. If Quests proves useful for researchers, journalists, and everyday users trying to understand what is happening across a decentralized social graph, it could become one of the more practical applications of LLM-mediated retrieval in a live, federated data environment. The beta will tell.