When your AI agents complete a code review, they should be able to sign it. That’s the premise behind Buzz, a new open-source workspace launched by Jack Dorsey’s Block on July 21, 2026.

Buzz merges team chat, Git hosting, and AI agent coordination into a single platform built on the Nostr protocol. Every message, workflow step, code review, and approval is a cryptographically signed event recorded in a shared log. Agents don’t just run in the background — they join channels as team members with their own keys, their own audit trails, and permissions you define.

The code is live on GitHub under the Apache 2.0 license. Teams can self-host or use Block’s managed service at buzz.xyz.

What’s Actually Different Here

Most teams today run AI agents in one place and collaborate in another — Slack for chat, GitHub for code, some agent platform for automation. Buzz collapses all of that into a single environment where humans and agents share the same timeline.

What makes that possible is the Nostr protocol. Nostr is a decentralized messaging protocol built around signed events and public/private key pairs. Traditionally used for social applications, Block adapted it to give AI agents their own cryptographic identity — meaning every action an agent takes is verifiable, auditable, and permanently attached to that agent’s key.

This is not a minor design choice. In enterprise settings, auditability matters. When an AI agent reviews a contract, approves a pull request, or executes a workflow, the question “who did this and when?” needs a reliable answer. With Buzz, that answer lives in the protocol itself, not in a log file that can be overwritten.

The platform is model-agnostic and agent-agnostic. Teams can deploy agents using Anthropic’s Claude Code, OpenAI’s Codex, or goose — the open-source agent framework Block released in January 2025. You’re not locked into any provider.

The Broader Shift This Signals

Block is not the first company to say AI agents should be first-class team members. But this is one of the first production-ready, open-source platforms that operationalizes it.

The traditional model treats AI as a tool you invoke and then review. You write the prompt, the AI produces output, a human approves it. That model breaks down as agents get more capable and more autonomous. When your AI agent is running for hours, working across multiple files, coordinating with other agents, and completing multi-step workflows — the tool metaphor stops making sense.

Buzz treats agents the way modern teams treat contractors: they have access, they do work, they leave signed artifacts behind. The transparency and accountability are built in, not bolted on.

For businesses already running AI agents in production — and there are more every week — that design philosophy matters. Shadow AI is one of the top concerns CIOs raised in every survey from the past twelve months. Agents that operate outside observable systems are an audit risk and a security risk. Buzz is a counterproposal to that problem.

What This Means for Business

The case for a platform like Buzz grows clearer when you look at what enterprise AI teams are actually trying to do. They want AI that completes work, not just assists with it. They want that work to be traceable. They don’t want to manage five separate systems to get there.

An open-source approach carries real advantages here. You can deploy Buzz on your own infrastructure, so sensitive data never leaves your environment. You can inspect the relay code, extend it for your specific workflows, and avoid vendor lock-in at a moment when the AI tools market is shifting fast. Free also matters in a budget environment where AI spend is already under scrutiny.

The limitations are real too. Buzz is still early. Mobile and desktop clients are in progress. The web-of-trust reputation features are on the roadmap, not shipped yet. Teams adopting it now are early adopters, which means tolerance for rough edges.

But the direction is sound. The future of enterprise AI isn’t agents running in isolation on behalf of humans. It’s humans and agents in the same room — same channels, same commit history, same accountability. Buzz is an early, credible attempt to build that environment in the open.

Enterprise DNA’s view: this is worth watching. As AI agents take on more operational work inside businesses — answering questions, running reports, handling customer interactions — the infrastructure that makes their work auditable and trustworthy becomes a competitive necessity, not a nice-to-have. Block is betting on open infrastructure as the answer. Given where the enterprise AI market is heading, that is a reasonable bet.