Back to List
Bluesky Team Unveils Attie: A New AI-Powered Assistant for Customizing Social Media Algorithms
Product LaunchBlueskyArtificial IntelligenceDecentralized Web

Bluesky Team Unveils Attie: A New AI-Powered Assistant for Customizing Social Media Algorithms

The team behind the decentralized social network Bluesky has introduced Attie, an innovative AI assistant designed to give users control over their social media feeds. Unveiled at the Atmosphere conference by former CEO Jay Graber and CTO Paul Frazee, Attie leverages Anthropic’s Claude AI and is built upon the AT Protocol (atproto). The application represents a significant shift in social media dynamics, moving away from platform-dictated discovery toward user-defined algorithms. By utilizing the open nature of the AT Protocol, Attie allows individuals to build and refine their own content curation systems, marking a new era of personalization in the decentralized web ecosystem.

The Verge
Share

Key Takeaways

  • User-Centric Algorithms: Attie is an AI assistant that empowers users to build and manage their own custom feed algorithms.
  • Powered by Anthropic: The application utilizes Claude, the advanced AI model from Anthropic, to facilitate feed customization.
  • Built on AT Protocol: The tool is integrated with Bluesky’s underlying decentralized framework, known as atproto.
  • Leadership Involvement: The project was unveiled by Bluesky’s former CEO Jay Graber and current CTO Paul Frazee at the Atmosphere conference.

In-Depth Analysis

The Shift to Algorithmic Choice

At the Atmosphere conference, the leadership behind Bluesky introduced Attie, a dedicated application designed to dismantle the traditional "black box" approach to social media feeds. Unlike mainstream platforms where algorithms are proprietary and opaque, Attie provides a transparent interface for users to dictate what they see. By allowing users to "build your own algorithm," the team is fulfilling a core promise of the decentralized web: giving power back to the individual user rather than the platform owner.

Technical Integration with Claude and AT Protocol

The technical foundation of Attie is a strategic combination of generative AI and decentralized protocols. By integrating Anthropic’s Claude, Attie provides a conversational or assistive layer that helps users navigate the complexities of content filtering and discovery. This is made possible by the AT Protocol (atproto), the open-source foundation of Bluesky. Because the protocol is designed for interoperability, Attie can interact with the network's data to reorganize and prioritize content based on the specific parameters set by the user, rather than following a singular, company-mandated engagement metric.

Industry Impact

The launch of Attie signals a major milestone for the decentralized social media (DeSoc) movement. By decoupling the user interface and the algorithm from the data layer, Bluesky is demonstrating a practical application of "composable" social media. For the broader AI industry, this highlights a growing trend of using Large Language Models (LLMs) not just for content generation, but as navigational tools for information architecture. If successful, Attie could set a precedent for other social platforms, forcing a shift toward algorithmic transparency and user agency across the digital landscape.

Frequently Asked Questions

Question: What is Attie and who created it?

Attie is an AI assistant designed for customizing social media feeds. It was developed by the team behind Bluesky and introduced by former CEO Jay Graber and CTO Paul Frazee.

Question: How does Attie use AI to change the user experience?

Attie is powered by Anthropic’s Claude AI. It allows users to build their own custom algorithms, giving them direct control over how content is filtered and displayed in their feeds, rather than relying on a platform-imposed algorithm.

Question: What technology serves as the foundation for Attie?

Attie is built on top of the AT Protocol (atproto), which is the underlying decentralized technology that powers the Bluesky social network.

Read Original Article

Related News

Evaluating the Efficiency of MiniMax Agent: An In-Depth Look at Architecture and Real-World API Performance
Product Launch

Evaluating the Efficiency of MiniMax Agent: An In-Depth Look at Architecture and Real-World API Performance

This analysis explores the practical utility of the MiniMax Agent, focusing on its internal architecture and its performance during real-world task execution. Based on a technical review by Shittu Olumide, the article delves into the specific components of the MiniMax ecosystem that were not addressed during its initial launch. By testing the agent directly against its actual API, the evaluation provides a transparent look at how the system handles functional requirements. The discussion highlights the importance of moving beyond marketing materials to understand the structural design and operational capabilities of AI agents. This deep dive aims to determine whether the MiniMax Agent truly simplifies professional workflows or if its architecture presents unique challenges for developers and end-users seeking to integrate it into their daily tasks.

Product Launch

ByteDance Launches Seedance 2.5: Revolutionizing AI Video with 30-Second Generations and Multimodal Referencing

ByteDance's Seed Team has officially unveiled Seedance 2.5, a next-generation video creation model designed to transition AI video from short clips to complete creative works. Building upon the unified multimodal audio-video architecture of Seedance 2.0, this update introduces the ability to generate high-quality 30-second clips in a single pass, with support for multi-round extensions to create multi-minute content. Key advancements include a massive upgrade to multimodal referencing—allowing up to 30 images and 10 video/audio clips as inputs—and timestamp-level editing for precise control. Seedance 2.5 focuses on foundational generation and flexible referencing to improve shot continuity, motion quality, and audiovisual consistency, marking a significant step forward in AI-driven storytelling and productivity.

QM: A New Multiplayer AI Agent Harness for Collaborative Startup Workflows in Slack and Web
Product Launch

QM: A New Multiplayer AI Agent Harness for Collaborative Startup Workflows in Slack and Web

QM is an innovative multiplayer agent harness designed specifically for the startup environment, bridging the gap between personal AI assistants and company-wide automation. Operating across Slack and web interfaces, QM provides isolated workspaces for individual employees while enabling seamless collaboration in shared channels and projects. The platform features scoped memory, granular permissions, and a durable sandbox for each user and room. Notably, QM is vendor-agnostic, allowing users to switch between models like Pi, OpenCode, Codex, and Claude Code. Its capabilities range from searching internal databases and triaging emails to managing GitHub repositories and deploying custom internal web apps, all powered by a headless core and Postgres-backed architecture. This tool aims to streamline startup operations by providing a flexible, secure, and highly customizable AI infrastructure.