Why Buzz May Be the First Truly AI-Native Workplace

On July 21, 2026, Jack Dorsey's Block released Buzz, a free, open-source collaboration platform that looks, at first glance, like yet another Slack clone. It has channels, threads, direct messages, voice huddles, and media sharing. Anyone who has lived inside a modern team chat tool will feel instantly at home. But the familiar surface conceals a genuinely different architecture, and that architecture is the story. Buzz was not built for teams of people who occasionally summon a chatbot. It was built for teams in which half the workforce is software.

The Technical Departure: Agents as First-Class Citizens

In Slack, Microsoft Teams, and virtually every incumbent platform, an AI assistant is a bolt-on. It lives behind a bot token, appears when invoked, and operates with permissions that are coarse, opaque, and administered by the platform vendor. Buzz inverts this model. Every participant, human or machine, receives its own cryptographic key pair on the decentralized Nostr protocol. An AI agent in Buzz has its own portable identity, its own channel memberships, and its own audit trail. You add an agent to a channel exactly the way you would add a coworker, and from that moment it can open code repositories, submit patches, review pull requests, run automated workflows, edit shared canvases, join voice huddles, and even coordinate other agents.

Three architectural choices make this technically significant.

First, Buzz is model-agnostic. It speaks the Agent Client Protocol, which means Anthropic's Claude Code, OpenAI's Codex, Block's own goose agent, or any compliant agent can operate inside the workspace on equal footing. No single AI vendor owns the room.

Second, Buzz is decentralized and self-sovereign. Built on Nostr and released under an Apache 2.0 license, it can be self-hosted and modified. Organizations wary of vendor lock-in, or of routing sensitive engineering conversations through a proprietary cloud, can run the entire stack themselves.

Third, Buzz collapses the toolchain. By bundling chat, Git hosting, code review, and workflow automation into one substrate, Block is betting that a single shared context beats the fragmented mix of Slack, GitHub, CI dashboards, and search indexes that most engineering organizations currently stitch together. When an agent and a human argue about a patch, the conversation, the diff, and the deployment all live in the same place, under the same identities, in the same auditable log.

The honest caveat is that the security model is unproven at scale. An agent in a channel can read everything in that channel, and a Slack employee raised exactly this data-leakage concern in the Hacker News discussion following launch. Block's answer is scoped identities and full audit trails rather than blanket bot tokens, but whether that holds up under real multi-tenant enterprise use remains to be demonstrated.

An Illustrative Example: Shipping a Hotfix at 2 a.m.

Imagine a mid-sized SaaS company whose payment service starts throwing errors overnight. In the traditional stack, the on-call engineer wakes up, reads a PagerDuty alert, opens Slack to coordinate, switches to GitHub to read code, opens Datadog for logs, writes a patch, waits for CI in yet another tab, and pings a colleague for review.

In a Buzz-native version of that night, the sequence looks different. A monitoring agent with its own cryptographic identity posts the anomaly into the #payments channel, attaching the relevant log excerpts. A triage agent, already a member of the channel, opens the repository, bisects the recent commits, and identifies a likely culprit. It drafts a patch and submits it for review inside the same channel. The on-call engineer, woken by a single notification, reads the agent's reasoning in-thread, requests one change by replying in plain language, and approves the revised patch. A deployment agent runs the workflow, and every action in the chain, from detection to deploy, is signed by the specific identity that performed it. In the morning, the postmortem is not a reconstruction from five tools; it is a scroll through one thread, with cryptographic attribution for every step. That auditability is the quiet killer feature, because it answers the question every enterprise will ask about agentic AI: who did what, and on whose authority?

How the Major AI Players Are Affected

The company most directly in the crosshairs is Salesforce, which paid $27.7 billion for Slack in 2021 and now finds its collaboration crown jewel challenged by a free, open-source alternative at a moment when its stock is already down more than 35 percent year to date. Slack still serves over 200,000 paid customers, including 77 of the Fortune 100, so the incumbency advantage is enormous, but Buzz attacks the exact territory, agent-and-human shared workflows, on which Salesforce has staked its AI strategy.

Microsoft is squeezed from two directions. Buzz positions itself as an alternative not only to Teams but also to GitHub, the $7.5 billion code-hosting platform that anchors Microsoft's developer ecosystem. If agent-driven development consolidates into unified workspaces, GitHub's role as the neutral home of code becomes contestable for the first time in years.

For Anthropic and OpenAI, Buzz is more opportunity than threat. Because the platform is model-agnostic, it functions as neutral distribution: Claude Code and Codex both gain a venue where they can act as full team members rather than embedded features of someone else's product. The competitive pressure shifts from owning the interface to being the best agent inside it.

Block itself is making a pointed bet. Having recently cut more than 4,000 jobs in an AI-driven restructuring, the company is essentially productizing its own thesis that future teams will be part human and part machine, and that the venue where they meet should be open rather than proprietary. Meanwhile, the broader race is already crowded: Paradigm's CTO has unveiled Centaur, a "virtual employee" that runs inside Slack, and open-source incumbents like Mattermost and Rocket.Chat will not cede the self-hosted market quietly.

The Bottom Line

Buzz will not displace Slack next quarter. Enterprises will rightly demand proof on administration, compliance, reliability, and the unresolved question of what an agent should be allowed to read. But the platform reframes the debate. The question is no longer whether AI belongs in the workplace chat window; it is whether the workplace itself will be a proprietary product or an open protocol. Dorsey has placed his bet, and for the first time since Slack redefined office communication a decade ago, the incumbents have to answer a question they did not write.

Sources: Block announcement and documentation as reported by The Next Web, The Register via Dealroom, Salesforce Ben, TMC Insight, and Northeast Times, July 2026.