Block, the technology company founded by Jack Dorsey, has open-sourced Berd, a desktop application designed to provide a unified environment for working with AI agents across different models, tools, and projects. Berd is now available on GitHub under the Apache 2.0 license, allowing users to freely use, modify, and redistribute the application on macOS, Windows, and Linux. The latest release, version 0.6.2, marks its seventh public iteration and includes contributions from 91 developers.
Berd is a locally installed graphical desktop application rather than a browser-based workspace. It is positioned as a "daily AI work surface" where users can initiate chats, attach files or folders, select agents and models, work within persistent projects, configure AI providers, manage skills and extensions, review session histories, and build automations. The design emphasizes visible operational states, distinguishing it from generic chatbot wrappers by making project, file, agent, model, provider, and session states transparent to users.
The impetus for Berd's development was a fragmented experience within Block, as employees worked with several AI agents like Block's Goose, Anthropic's Claude Code, and OpenAI's Codex. Berd provides a consistent desktop application that unifies these disparate interfaces and configuration systems, enabling users to manage context more effectively. Unlike typical enterprise AI products that treat agents as interchangeable chat windows, Berd assigns each agent a distinct role, instructions, skills, tools, and visual identity. This includes animated characters known as "Gloopies," which serve as recognizable avatars for different agent functionalities.
Berd targets enterprises and teams that require a unified platform for managing AI agents across various projects and tools. It is designed to make agentic work accessible beyond engineering fields, enabling users to begin with a conversation and gradually incorporate tools, context, and structure as needed.
Work implications: Berd could streamline workflows for teams working with multiple AI models, reducing the need to switch between different interfaces and systems, thus enhancing productivity in multi-agent environments.
Originally reported by VentureBeat