DeepHorus is an innovative tech media and intelligence platform focused on AI, blockchain, startups, and emerging technologies.
Published on

Block’s new Apache 2.0 agent workspace Berd works across models and harnesses, stores conversation history locally

Authors
  • avatar
    Name
    DeepHorus
    Twitter

Block, the financial technology company founded by Jack Dorsey, has open-sourced Berd, a desktop application designed as a single workspace for AI agents across different models and tools. The app was originally built for Block's own employees, who struggled to manage multiple AI assistants. Now it is available on GitHub under the permissive Apache 2.0 license for anyone to use, modify, or redistribute commercially.

Berd is a locally installed graphical desktop application rather than a browser-based workspace. The repository reached version 0.6.2 on August 18, with 91 contributors, and provides free downloads for macOS, Windows, and Linux. Block calls Berd a 'daily AI work surface' where users can attach files, select agents and models, manage projects, and build automations.

A core design principle is transparency. According to Berd's specification, users should always be able to see which project, files, agent, model, provider, and session state are active in a conversation. Block believes this visible operational state separates Berd from generic chatbot wrappers, and the interface deliberately exposes failures, unavailable providers, and streaming states.

Berd emerged from a real problem inside Block. Employees used capable agents like Block's own Goose, Anthropic's Claude Code, and OpenAI's Codex, but working across them meant juggling different interfaces and configuration systems. Lucinda Bell, Block's technology communications lead, said Berd gives teams 'one consistent desktop application across models and harnesses.'

For enterprises, Berd's persistent projects and agent personas may reduce context-switching and make agentic work accessible beyond engineering. The app features animated characters called 'Gloopies,' with predefined roles like Pushback, Choosey, and Copycat, each tailored to different work styles. This visual layer is not a replacement for personal interaction but helps users distinguish between configured agents.

Block does not see Berd as a mobile product; its open-source platform Buzz provides the mobile experience. By releasing Berd under Apache 2.0, Block invites community contributions and positions the application as a potential standard for managing multi-agent workflows. Future development will likely focus on expanding its capabilities and integrations.

Read original article →Source: venturebeat