What is Buzz? Jack Dorsey’s Open-Source Workspace Where Humans and AI Agents Work Together

Published for students and developers interested in the future of AI collaboration

In the last year, AI tools have moved from simple chatbots to powerful agents that can write code, research topics, edit documents, and even talk to other agents. The big problem has been that these agents usually work alone. They do not share a common workspace with humans or with each other. Jack Dorsey’s company Block (the same company behind Square and Cash App) has released an open-source solution called Buzz.

Buzz is a free, open-source collaboration platform that works like Slack, but it is built specifically for teams of humans and AI agents. It already has tens of thousands of stars on GitHub and is one of the most interesting open-source AI projects of 2026.

What Exactly is Buzz?

Buzz is not an AI model. It does not generate answers by itself. Instead, it is a workspace (community) where:

• Humans join with their own identity.
• AI agents join as full teammates (not just bots).
• Agents can chat with each other, delegate tasks, review code, and work on projects together.
• Everything is logged and cryptographically signed so there is a clear audit trail.

Think of it as “Agent Slack”. You can create channels, invite people and agents, and work together on coding, content creation, research, or any other task.

The official website is buzz.xyz and the source code is available on GitHub under the Apache-2.0 license (github.com/block/buzz).

How the Architecture Works

Understanding the architecture is important because Buzz is different from normal Slack bots.

1. Desktop / Mobile App
   You download the Buzz app on your computer (Windows, Mac, Linux) or phone. This is where you chat.

2. Buzz Relay (the community host)
   The actual workspace (community) runs on a server. This “relay” brokers messages between humans and agents. It also stores data in Postgres, uses Redis, and can host Git repositories.

3. Bring Your Own Agent
   The agents do not come from Buzz. You connect agents that already run on your computer or server, such as:
   • Claude Code (Anthropic)
   • OpenAI Codex
   • Block’s own Goose
   • Cursor
   • Local models via LM Studio
   • And others that support the Agent Client Protocol

Each agent has its own cryptographic identity (public/private key pair) and is tied to a human owner. The human can bring one or many of their agents into the community. Agents from different people can also work together.

1. Security and Audit
   Every action is signed with cryptographic keys. This creates a permanent, verifiable history of who (human or agent) did what. It is similar in spirit to how blockchains keep records, but without needing a public blockchain.

2. Git Integration
   Buzz has native support for Git repositories and worktrees. Multiple agents can work on the same code at the same time without overwriting each other’s changes.

Why Students and Developers Should Care

For engineering and diploma students in Andhra Pradesh and across India, Buzz is useful for several reasons:

• Learn multi-agent systems practically: Instead of only reading about AI agents, you can set up a researcher agent, a writer agent, and an editor agent and watch them collaborate.
• Open-source contribution and self-hosting: You can run the entire community on a cheap VPS (Virtual Private Server) and own your data.
• Future workplace skills: Companies are moving toward “humans + agents” teams. Experience with tools like Buzz, Claude Code, Codex, and Git-based agent workflows will be valuable.
• Coding projects: Use it for college projects where agents help with documentation, testing, or feature implementation while you review and guide them.
• No vendor lock-in: Because it is open source and model-agnostic, you can switch between different AI providers.

High-Level Setup Overview

1. Download the Buzz desktop app from buzz.xyz and create your identity (save the private key securely, like a seed phrase).
2. Set up or join a community. For full control and lower cost, host the relay yourself on a VPS (Hostinger and other providers offer easy options; Docker is used).
3. Connect your preferred agent harnesses (Claude Code, Codex, Goose, etc.).
4. Create channels, agents, and even teams of agents.
5. Tag agents in messages to start workflows. Agents can then talk to each other and hand off tasks.

Mobile apps are also available so you can monitor and chat from your phone.

Real-World Example

One practical use case is content or project work. You can create:

• A Researcher agent that finds trending topics or ideas.
• A Writer agent that drafts content or code.
• An Editor agent that reviews for quality, clarity, and correctness.

Once you give the initial instruction, the agents can collaborate in the channel, update each other, and produce the final output. You only step in when needed.

Conclusion

Buzz represents a shift from “chat with one AI” to “work with a team of AI agents in a shared, auditable workspace”. It is open source, self-hostable, and designed for real collaboration between humans and agents.

For students learning AI, software engineering, and future workplace tools, exploring Buzz is a practical way to understand how multi-agent systems will be used in companies. Start by downloading the app, connecting one agent you already use, and experimenting with a small project.