Jack Dorsey announced Buzz, an open-source group chat platform built by Block, on July 21, 2026. The product positions itself as a challenger to Slack and GitHub, with a defining design choice: human users and their AI agents share the same conversations rather than operating through separate channels or API integrations. (TechCrunch)

Dorsey described Buzz on X as "model-agnostic, decentralized, self-sovereign, and open source." His X/Twitter profile bio further characterizes it as "an open-source workspace for humans and AI agents, on a relay you own." (TechCrunch)

The codebase has been uploaded to GitHub, and a free desktop application is available for macOS, Windows, and Linux. The product's website is hosted at buzz.xyz. (TechCrunch)

Buzz is being candid about its maturity. The project acknowledges it is in its "early stages" and is not yet ready for teams to fully migrate from existing platforms. That puts it closer to an early public build than a production-ready Slack replacement on day one. (TechCrunch)

Block, the company behind Buzz, is best known for Square, Cash App, Afterpay, and Tidal. Buzz represents a departure from that portfolio's fintech and commerce focus into developer-facing collaboration infrastructure. (TechCrunch)

The competitive landscape Dorsey is entering already has entrants thinking along similar architectural lines. Paradigm partner and CTO Georgios Konstantopoulos recently unveiled Centaur, an open-source product described as a "virtual employee" that runs inside Slack or via an API. The overlap is clear: both projects assume that AI agents will be first-class participants in team workflows, not background tools invoked through separate interfaces. (TechCrunch)

The architectural choices Dorsey has flagged warrant attention. "Model-agnostic" means the platform does not lock users into a single LLM provider. "Decentralized" and "self-sovereign," combined with the "relay you own" framing in his bio, suggest an architecture where message routing and data custody sit with the user or their organization rather than with a central platform operator. For teams evaluating where their conversation data lives and who controls retention, those are structurally different promises than what Slack or Microsoft Teams offer today.

The open-source approach also matters for adoption dynamics. A GitHub-available codebase lets security teams audit the product before deployment, lets infrastructure teams self-host if the architecture supports it, and lets developers extend the platform through contributions rather than vendor-supported plugin marketplaces. Whether that translates into a viable ecosystem depends on how actively Block maintains the repository and whether third-party contributors engage.

Dorsey's competitive positioning spans two distinct product categories. Slack owns synchronous team communication; GitHub owns asynchronous code collaboration and, increasingly, broader project management. A single platform attempting to address both, while also introducing agent-native chat as a core primitive rather than a bolt-on integration, is an ambitious scope. The early-stages caveat suggests Block is aware of the gap between vision and current capability.

What is genuinely new here is not group chat, open-source collaboration tools, or AI agents operating in enterprise environments. It is the assertion that all three belong in the same conversation layer, with the user owning the relay. If that architecture proves out, the implications go beyond competitive displacement of existing tools. They touch on how teams govern agent behavior, audit agent-initiated messages, and maintain data sovereignty in workflows that are increasingly mediated by models none of the participants fully control.

The project is available now for evaluation. Whether teams adopt it at scale will depend on factors Buzz has not yet demonstrated: enterprise-grade reliability, a mature access-control model, and the kind of integration ecosystem that took Slack years to build. The code is on GitHub, the desktop apps are free, and Dorsey has set the architectural terms clearly. The rest is execution.