How should modern teams work alongside their AI counterparts? Jack Dorsey’s Block is attempting to answer that question with Buzz, a workplace collaboration app launched in July 2026 that places human employees and AI agents in the same shared workspace. Buzz is framed as a group chat platform for teams of people and agents of all sizes, designed to reduce dependence on incumbent tools such as Slack for messaging and GitHub for code-centric collaboration. Rather than treat AI as a bolt-on feature, the product positions agents as first-class participants in everyday conversations, projects, and review cycles.
At its core, Buzz is an open-source workplace chat platform and collaboration environment tailored to modern software development. It combines familiar communication primitives—channels, threads, and direct messages—with voice calls and media sharing for images, audio, video, and other files required in engineering and product work. Teams keep discussion and assets inside a single interface instead of juggling disconnected apps. Integrated code repositories appear directly within the workspace, allowing developers to move from conversation to commits, reviews, and merges in a continuous flow.
An open-source chat workspace where conversation, code, and collaboration flow through a single continuous interface
Buzz’s workflow system links chat to repositories, turning coordination into automated pipelines. Teams can define workflows that react to messages, code events, or agent actions, using the same space to trigger builds, tests, deployments, or custom processes. This event-driven design is meant to collapse the distance between talking about work and executing it, particularly in environments where both humans and AI systems are contributing changes. Project coordination, code review, and operational tasks become facets of one shared log of activity.
A central claim of Buzz is that humans and AI agents can operate in the same rooms and threads with a common audit trail. Third-party agents are able to join channels, participate in discussions, and run workflows alongside human teammates, whether they specialize in coding, analysis, or operations. The platform is described as model-agnostic, so organizations can integrate multiple language models or agent frameworks rather than commit to a single provider. In practice, this is meant to let teams pair different AI capabilities with specific projects while retaining a unified identity and history for every participant.
Technically, Buzz is built on the Nostr protocol, using a decentralized, event-based communication layer. Messages, code commits, workflow steps, and agent invocations share the same event bus, producing an audit trail across chat and repositories. Under the hood, Buzz’s proprietary event bus is engineered to process 50,000 events per second while enforcing a permission model that spans chat, code, and AI agents. Block emphasizes that Buzz runs “on a relay you own,” aligning the product with self-sovereign collaboration in which organizations control their own infrastructure and data.
The software is open source and released under the Apache-2.0 license, enabling self-hosting, customization, and independent instances that operate without reliance on centralized SaaS vendors. In doing so, Buzz presents itself as a challenger to established collaboration platforms and an experiment in how humans and AI agents might share a workplace.
