Jack Dorsey has spent years betting on open protocols over proprietary platforms, and the Block Buzz platform is his most direct challenge yet to the tools that currently run how software teams work. Launched on July 21, 2026, Buzz is a free, open source workspace where employees and AI agents occupy the same channels, share code repositories, and hold their own cryptographic identities — all running on the Nostr protocol.
Summary
Key takeaways
- Block launched Buzz, a free and open source platform built on the Nostr protocol, positioning it as a direct alternative to Slack and GitHub for human-AI team collaboration.
- Every participant — human or AI agent — receives a platform-independent cryptographic keypair, meaning identity and history travel with the holder, not the platform.
- Buzz supports agents powered by Anthropic’s Claude Code, OpenAI’s Codex, and Block’s own Goose framework, making it model-agnostic from day one.
- The code is published under an Apache 2.0 license and available at github.com/block/buzz; teams can self-host or use Block’s managed version.
- The platform ships at version 0.4.21 with desktop apps for macOS, Windows, and Linux; mobile clients remain unfinished.
One workspace to replace several
The question Buzz answers is one that every engineering team is quietly wrestling with: when AI agents are doing real work alongside humans, where do they actually live? Right now the answer is scattered across chat platforms, code hosts, ticketing systems, and CI dashboards — tools that weren’t built with agents in mind.
Buzz folds all of that into a single interface. It includes channels, threads, direct messages, voice, media sharing, code repositories, canvases, search, an audit log, and automated workflows. Jack Dorsey announced the launch on X, describing it as “model-agnostic, decentralized, self-sovereign, and open source,” and explicitly framed it as a move to reduce Block’s own dependence on Slack and GitHub.
That internal origin matters. Block built Buzz itself using agents, according to Thomas Petersen, a principal designer and AI strategist at the company. Petersen said the entire platform was developed using agents, with the interface “more sculpted over time than designed up front.” The product isn’t a theoretical answer to a hypothetical problem — it’s what Block actually needed and couldn’t find elsewhere.
Core features and integrated tools
In practical terms, Buzz looks familiar to anyone who has used a modern team chat tool. The difference shows up in what agents can do inside it. Rather than functioning as bots that respond only when summoned, AI agents on Buzz participate as independent workspace members with their own permissions and activity history. They can post messages, review code, run approved automations, create their own channels, and update their own memories.
Lucas Isaza, who works on AI and product at Block, argued that tools like Asana, Jira, and Linear didn’t just facilitate work — they shaped how teams organized themselves, baking sprints and Kanban into company culture. A configurable workspace like Buzz removes those constraints. In his framing, a single thread can contain the diagnosis, the task, the fix, and the surrounding conversation, collapsing what previously spanned several tools into one record of why a decision was made.
Support for multi-model AI agents
Buzz is explicitly model-agnostic. Teams can deploy agents powered by Anthropic’s Claude Code, OpenAI’s Codex, or Block’s own open source Goose framework. Companies can also connect internal databases, codebases, and file systems, with granular control over what each agent can access. Users set channel- or thread-specific rules for every agent, keeping human oversight in the loop.
Why the Nostr protocol changes the identity problem
The choice of Nostr as the foundation isn’t aesthetic — it solves what Block calls the most fundamental problem in multi-agent work: identity. On most platforms, an account belongs to the platform. Revoke access, switch tools, or get locked out, and the agent’s history goes with it.
On Buzz, every participant holds a cryptographic keypair that belongs to them rather than to the platform. That applies to humans and agents equally. An agent’s identity, history, and reputation can travel across any Nostr-compatible system. Every message, reaction, workflow step, code review, and git event is recorded as a single signed event, creating what the project’s repository describes as a hash-chain audit trail — whether the author is a person or an automated process.
Platform-independent cryptographic identities for users and agents
This approach has a meaningful implication beyond portability. When an AI agent has a persistent, verifiable identity, teams can assign it accountability. Audit logs become legible records of who — or what — approved a change, merged a branch, or triggered a workflow. That’s not just a technical nicety; it’s the kind of traceability that enterprise teams and compliance functions will eventually demand from any AI-integrated workflow.
Single relay as the source of truth and audit logging
Each Buzz workspace currently runs on a single relay, which the project describes as its single source of truth for events. That design choice keeps the system coherent and auditable, though it also means the decentralization story has practical limits at the workspace level — a tension worth watching as the platform matures and adoption scales.
Collaborative workflows and configurable AI agent interactions
“Every company is going to need a place where humans and agents work together,” said Bradley Axen, Block’s head of AI capabilities, in a statement. “The question is whether that place is proprietary or open.”
That framing gets at what makes Buzz strategically significant beyond its feature list. The enterprise software market has long been dominated by platforms that capture teams through lock-in — proprietary data formats, closed APIs, and integrations that only work within the vendor’s ecosystem. An open source, self-hostable alternative built on a public protocol sits in direct opposition to that model.
Isaza described experimenting with agents that generate a shared whiteboard for every thread, producing “better context handoffs with no additional work.” Block has tested the approach internally for months and, in his words, has “yet to figure out” the model that extracts the most from Buzz — an honest admission for a company launching a product it expects others to adopt.
Open source licensing, platform maturity, and strategic vision
Buzz is available now at version 0.4.21, with desktop builds for macOS, Windows, and Linux. Mobile clients are still under development, as are approval gates and additional voice features. The current release covers channels, threads, direct messages, canvases, media sharing, search, an audit log, desktop apps, automated workflows, and early Git integration.
Apache 2.0 license and deployment options
The source code is published under an Apache 2.0 license, giving teams the ability to self-host on their own infrastructure — controlling their own data, relay, and agents — or use Block’s managed version. That flexibility is meaningful for organizations with data residency requirements or security policies that preclude third-party hosting.
Block’s broader push for open protocols and decentralization
Buzz sits inside a broader pattern. Since departing Twitter, Dorsey has backed a string of tools built around protocols no single company controls. Nostr has been the common thread — Block has built or funded multiple projects on the relay-based protocol, including Bitchat, its Bluetooth mesh messenger. Buzz extends that philosophy directly into the workplace, applying the same decentralized identity logic to the tools people use to ship software every day.
Buzz isn’t the only product taking this direction. As TechCrunch reported, Paradigm partner Georgios Konstantopoulos recently open-sourced a similar product called Centaur, described as a virtual employee that runs inside Slack or via an API. The convergence of multiple teams on the same problem — agents that participate meaningfully in team workflows — suggests the category is forming fast, with open source as the early competitive frame.
What remains unresolved is whether teams beyond Block’s own engineers will find the paradigm shift manageable. Convincing organizations to rethink how humans and agents share accountability — not just add a bot to a channel — is a harder sell than any feature comparison. The companies that figure it out first will have built something genuinely new; Buzz is Block’s opening argument for what that looks like.
FAQ
What is Buzz and who developed it?
Buzz is an open source platform for teams where humans and AI agents collaborate in the same workspace. It was developed by Block, the payments company led by Jack Dorsey.
How does Buzz handle identity for users and AI agents?
Buzz uses the Nostr protocol, giving each user and agent a platform-independent cryptographic keypair as their identity. This means identity and history belong to the holder, not the platform, and can travel across any Nostr-compatible system.
Which AI models can agents on Buzz use?
Buzz supports AI agents powered by Anthropic’s Claude Code, OpenAI’s Codex, and Block’s own Goose framework. The platform is model-agnostic, allowing teams to connect agents built on any supported LLM or harness.
Is Buzz ready for mobile devices and how mature is the platform?
As of version 0.4.21, Buzz has desktop apps for macOS, Windows, and Linux. Mobile clients, approval gates, and additional voice features are still under development, and the platform self-describes as being in its early stages.
Article produced with the assistance of artificial intelligence and reviewed by the editorial team.
