From Chat Tool to Shared Brain: What Buzz Actually Is
Buzz is an open source collaboration tool that combines team chat, code repositories and automated workflows in a single workspace where human teammates and AI agents share channels, threads and identities as equal participants, built on the decentralized Nostr messaging protocol to support secure, encrypted human-machine teamwork at scale.
Block has launched Buzz as a free, open-source workspace that looks at first glance like Slack or Discord, with channels, threads, DMs, voice, media sharing and code repositories. The difference is philosophical: Buzz treats AI agents as full participants in these conversations, not as quiet sidekicks you consult in another window. Each agent can join channels, reply in threads and be mentioned like any teammate, which changes AI agent collaboration from a private, copy‑paste interaction into a shared, inspectable dialogue. That is the key shift: instead of hiding your work with agents, Buzz drags it into the team’s main room, where decisions are made, challenged and audited. It is not “AI in chat”; it is a shared brain for humans and machines.
Passport Identities: Why Crypto Keys Matter for Human-Machine Teams
Buzz’s most provocative idea is that every participant—human or agent—gets a passport-style identity backed by cryptography. On Nostr, every participant is, at its core, a public‑private keypair, and Buzz gives each agent its own keypair and treats it as an independent actor. Then it adds a second signature that cryptographically ties that agent back to the person who owns it, creating what Block describes as a cryptographic paper trail that others can review.
This matters because true human-machine teamwork demands accountability. If an AI merges code or posts advice, teams need to know which human stands behind that automation. With this identity scheme, a workspace can enforce rules like allowing only agents owned by members into private channels or gating sensitive workflows. Buzz’s design turns AI agents from anonymous bots into traceable team members with linked owners, and Block has proposed this identity system as an upstream Nostr extension, even though it has not been accepted yet. That is a strong signal: Buzz is not content to sit on top of existing protocols; it wants to redefine what identity means in collaborative systems.
One Workspace for Code, Chat and Agents: A Slack Alternative with Teeth
On the surface, Buzz is a Buzz workspace platform that pulls together team chat, code repositories and automated workflows in one place. Underneath, it wires in AI coding agents through the Agent Client Protocol (ACP), the open standard originally introduced to connect coding agents into tools. Buzz can work with any model and popular agent framework, so development teams can plug in Claude Code, OpenAI Codex, Block’s own goose or their own agents.
This is where Buzz takes a direct swing at today’s stack of Slack plus Git-based tools. Jack Dorsey has said the platform aims to reduce the company’s dependence on Slack and GitHub. Instead of hopping between chat, repos and CI dashboards, teams can build automated workflows where agents respond in-channel, push commits and coordinate work in one shared space. Buzz’s code has been on GitHub under an Apache 2.0 license since earlier this year, but the public launch coincides with Block hosting its own Nostr relay and opening signups. It already supports desktop clients on macOS, Windows and Linux, with mobile still in development. For teams that want AI agent collaboration built into their daily tools rather than sprinkled on top, that consolidation is compelling.
Control the Pipes or Click and Go: Infrastructure and Open Source Power
Because Buzz runs on Nostr, it needs a relay to function, and here Block makes an opinionated bet on openness. Any team can run its own Nostr relay and keep full control of its infrastructure, with no traffic routing through Block at all. That design turns Buzz into something closer to a Discord server where no single company owns the core, aligning with its framing as an open source collaboration tool rather than a locked SaaS. For teams that do not want to manage this complexity, Block now offers hosted relays that are free at launch and framed as a beta.
This dual model—self-hosted or hosted—matters more than it might appear. If Buzz is to become the nervous system for human-machine teamwork, organizations will demand control over data flow and governance. Letting them own the relay while still providing an off‑the‑shelf option is a practical compromise. Meanwhile, the Agent Client Protocol is on the radar of the Linux Foundation’s Agentic AI Foundation steering committee, on which Block’s Bradley Axen sits, with the group considering adopting ACP. That points toward a broader ecosystem of compatible agent tools, not a closed Buzz island.
From Tools to Teammates: Why Buzz Signals a New Way of Working
Buzz does not appear in a vacuum. Block is reshaping itself around AI, having cut more than 40 percent of its staff—from over 10,000 employees to just under 6,000—with Dorsey pointing to AI as enabling “a new way of working.” Internally, the company’s BuilderBot already runs over 200,000 operations a day, merges about 1,500 pull requests a week and accounts for roughly 15 percent of production code changes. That lived experience is visible in Buzz’s design: it assumes agents will be constant collaborators, not occasional helpers.
In practice, Buzz is still young; its repository has only a bit over 100 stars so far, far fewer than goose. Block itself still uses Slack, and even its leaders admit Buzz may live alongside legacy tools for some time. Yet the direction is clear. Traditional chat platforms treat AI as bots bolted onto human workflows. Buzz treats AI as team members with identities, permissions and shared history. As mobile support arrives and the identity scheme matures, Buzz could pressure incumbents to rethink how AI is woven into collaboration. The real question is not whether Buzz replaces Slack, but whether teams are ready to let machines sit beside them in the channel, leaving a permanent, inspectable trail of how work gets done.
