Buzz Jack Dorsey: The Open-Source Workspace Uniting Humans, AI Agents, and Code

Jack Dorsey, co-founder of Twitter (now X) and Block (the company behind Square, Cash App, and other fintech products), has introduced a new platform called Buzz. Announced in July 2026, Buzz is positioned as a free, open-source collaboration workspace designed for teams of people and AI agents. It aims to reduce reliance on tools like Slack for messaging and GitHub for code hosting by bringing chat, project management, Git workflows, and agent collaboration into one shared environment.

Built by Block and available at buzz.xyz, Buzz is model-agnostic, decentralized, self-sovereign, and open source under the Apache 2.0 license. Desktop apps are available for macOS, Windows, and Linux, with the source code hosted on GitHub at github.com/block/buzz. The platform treats AI agents as full participants rather than simple chatbots, giving them cryptographic identities so they can post messages, review code, open pull requests, and collaborate alongside humans in the same channels and threads.

This article explains what the Buzz application is, how it integrates with GitHub-style workflows and Git hosting, and the role of the Nostr protocol that underpins its design.

What Is the Buzz Application?

Buzz is a native workspace created for the era of AI agents. Traditional team tools separate human conversation (Slack or Microsoft Teams), code and version control (GitHub or GitLab), task tracking, and AI assistance (separate chat interfaces or IDE plugins). This fragmentation forces people to copy context between tools, lose history, and treat agents as external helpers rather than team members. Buzz addresses that gap by putting humans and agents in the same rooms with shared context.

According to Block’s announcement, the most productive work happens when humans and agents occupy the same space, working on the same problems with the same information. Buzz provides channels, threads, direct messages, media sharing, voice capabilities, and automated workflows. Its interface will feel familiar to anyone who has used modern team chat tools, yet it is purpose-built for multi-agent collaboration.

Key characteristics include:

• Agent-first design: AI agents receive their own cryptographic identities and defined permissions. They can participate in conversations, post updates, review code, run approved automations, and build on each other’s work rather than operating in isolation.
• Model- and agent-agnostic: Teams can use agents powered by any large language model or agent framework, including Anthropic’s Claude Code, OpenAI’s Codex, Block’s own goose, or custom agents they build or bring themselves. The platform does not lock users into a single vendor’s models.
• Open source and self-hostable: The full codebase is public. Organizations can run their own instance on their own infrastructure for complete control over data, relays, and agents, or use Block’s hosted version at buzz.xyz.
• Unified workspace: Chat, planning, project management, coding, and pull requests live in one place. Discussion can flow directly into plans, code changes, reviews, and merges without constant context switching.

Buzz is still in its early stages, as noted by both Block and independent coverage. Early features emphasize shared rooms where specialized agents can compare notes, divide work, and contribute alongside people. Upcoming or in-progress capabilities include deeper Git project management so that open-source communities or internal teams can host repositories, discuss ideas, review changes, and let agents help write, test, and maintain code - all within the same environment.

For teams already relying on AI agents for coding, research, writing, or workflow automation, Buzz offers a single searchable record of human and agent activity. Every message, reaction, workflow step, review approval, and Git-related event can be treated as a signed, verifiable action. This creates transparency and auditability that proprietary platforms often lack.

The platform’s launch reflects broader industry movement toward agent-native tools. As companies integrate more autonomous or semi-autonomous AI systems, the need for a neutral, open place where those systems can collaborate with humans becomes clearer. Buzz attempts to meet that need without requiring teams to abandon their preferred models or frameworks.

How Buzz Integrates with GitHub and Git Workflows

One of Buzz’s explicit goals is to reduce dependency on separate tools for chat and code. Jack Dorsey’s announcement highlighted that the platform was built to lessen reliance on both Slack and GitHub. While full Git hosting inside Buzz was described as early or emerging at launch, the architecture already supports Git-related events and workflows in meaningful ways.

Buzz treats Git activity as first-class events within its event log. Because the system is built on signed events, commits, pull requests, reviews, and related actions can appear alongside chat messages and agent contributions in the same timeline. This creates a unified history instead of forcing developers to jump between a chat app and a separate code hosting service.

Practical integration points include:

• Git project management inside the workspace: Teams can move from discussion to planning, coding, reviews, and PRs without leaving the Buzz interface. Agents can participate in code review, suggest changes, or even generate and commit code under controlled permissions.
• Cryptographic signing of Git actions: Tools and helpers such as git-sign-nostr and related credential helpers allow Git operations to be signed with the same Nostr-based identities used for chat. This links code contributions to the same portable identity that an agent or human uses elsewhere in the system.
• Self-hosted or connected repositories: Organizations running their own Buzz relay can connect existing codebases or host repositories within the platform. Agents and humans then interact with that code in the shared channels where the conversation already lives.
• Workflow automation around code: Automated steps - tests, reviews, approvals, or deployments - can be expressed as signed events. Agents with appropriate permissions can trigger or complete those steps while remaining visible to the rest of the team.

Because the source code of Buzz itself lives on GitHub (github.com/block/buzz), developers can inspect the architecture, contribute improvements, or fork the project. The repository includes documentation on architecture, Nostr usage, agent skills, deployment, and more. This transparency lets teams understand exactly how Git events are handled and how to extend the system for their own needs.

For developers evaluating the platform, the combination of chat and Git in one place reduces the friction of context loss. An agent that has been discussing a feature in a channel can, with proper permissions, open a pull request or comment on code in the same continuous record. Humans can review both the conversation history and the code changes without switching applications. Over time, as Git hosting matures inside Buzz, open-source communities may find it useful as a self-sovereign alternative or complement to centralized hosts.

Integration is intentionally flexible rather than mandatory. Teams can continue using external GitHub repositories while bringing conversation and agent activity into Buzz, or they can experiment with deeper embedding of code workflows. The open-source nature means custom bridges and connectors can be built as needed.

Understanding the Nostr Protocol Behind Buzz

The technical foundation of Buzz is the Nostr protocol. Nostr stands for “Notes and Other Stuff Transmitted by Relays.” It is an open, decentralized communication protocol designed for censorship resistance and user sovereignty. Created in 2020 by a developer known as fiatjaf, Nostr was a response to concerns about centralized moderation and control on platforms such as Twitter.

Nostr is not itself a social network or an application. It is a simple protocol that defines how clients and relays exchange signed JSON events. Users (or agents) are identified by cryptographic keypairs—public keys that serve as persistent identities and private keys used to sign events. Anyone can run a relay; clients publish events to one or more relays and subscribe to events from them. There is no central server that owns the user’s identity or data.

Jack Dorsey has been a prominent supporter of Nostr for years. He has donated significant resources to its development, including Bitcoin and cash contributions, and has publicly advocated for open protocols that return data ownership to users. Buzz represents a practical application of those principles to the workplace and multi-agent collaboration.

In Buzz, every participant—human or AI agent—holds a cryptographic keypair. This identity is not tied to a platform account or an API key controlled by a vendor. It is portable and verifiable. An agent’s identity, history, and reputation can travel with it across any Nostr-compatible system. This solves a core problem in multi-agent environments: establishing trustworthy, independent identities that are not locked into a single proprietary platform.

Key properties of Nostr that matter for Buzz include:

1. Signed events: Every message, reaction, workflow step, review, or Git-related action is a cryptographically signed event. The signature proves origin and integrity.
2. Relay model: Buzz operates as a Nostr relay (or set of relays). Teams can use Block’s hosted infrastructure or run their own for full control. Data does not have to leave the organization’s chosen infrastructure.
3. Extensibility through NIPs: Nostr Implementation Possibilities (NIPs) define optional extensions. Buzz builds on the core NIP-01 event format and adds the kinds of events needed for team chat, agents, and code workflows.
4. Self-sovereignty: Because identity lives in the keypair rather than a central database, users and agents retain control. This aligns with the broader goals of reducing vendor lock-in.

Nostr has been used for microblogging clients (such as Damus and Primal), private messaging, and other applications. Its simplicity - clients are relatively smart, relays are relatively dumb - makes it adaptable. Spam and abuse remain challenges in open networks, and mitigation often relies on web-of-trust approaches, proof-of-work, or client-side filtering. In a team or enterprise deployment of Buzz, administrators control the relay and can set permissions, making the environment more contained than a fully public social network.

By choosing Nostr, Block and Jack Dorsey positioned Buzz as more than another chat tool. The protocol provides a neutral, open substrate for identity and events. Agents become first-class citizens with portable credentials. Teams gain the option of true self-hosting. And the same identity system can, in principle, interoperate with other Nostr-based tools in the future.

You may also like : The Ultimate Guide to AI Coding Tools 2026: Transforming Developer Productivity

Why Buzz Matters and What Comes Next

Buzz Jack Dorsey’s launch arrives at a moment when organizations are rapidly adopting AI agents for coding, research, customer support, and internal automation. Existing tools were not designed for agents to sit as equals in the same conversation as people. Proprietary platforms create fragmentation and dependency. Open-source alternatives that treat agents as full members with verifiable identities are still rare.

The combination of familiar chat patterns, emerging Git integration, and a Nostr foundation gives Buzz a distinctive profile. It is free to use in hosted form, fully open source for those who want to self-host, and deliberately model-agnostic so teams can choose the AI systems that fit their needs. Early adopters will need to evaluate its maturity, security model, and fit with existing workflows, but the architectural choices are clear and consistent with Dorsey’s long-standing interest in open protocols and user sovereignty.

For developers, the public GitHub repository offers a direct way to explore the code, experiment with local relays, and contribute. For teams, the hosted experience at buzz.xyz provides a low-friction entry point to test human-agent collaboration. For the broader ecosystem, Buzz demonstrates one concrete path toward workplaces where people and AI systems share context, identity, and responsibility in a transparent, auditable way.

As Git features mature and more agent frameworks gain native support, the platform has the potential to become a practical operating environment for agent-augmented teams. Whether it displaces existing tools or complements them will depend on real-world adoption, reliability, and the community that forms around the open-source project. What is already evident is that Buzz represents a deliberate attempt to build collaboration infrastructure that is open, portable, and designed for the multi-agent future rather than retrofitted onto older models of work.

In summary, Buzz is Jack Dorsey’s and Block’s answer to the question of where humans and AI agents should work together. It unifies conversation and code under cryptographic identities provided by the Nostr protocol, offers self-hosting and hosted options, and remains open for inspection and extension. Teams exploring the next generation of collaboration tools now have a concrete, inspectable project to evaluate.