Block Launches Buzz: A Decentralized Team Collaboration Platform for AI Agents

The real question isn't whether it can replace Slack, but whether it forces Slack, Teams, Notion, and others to answer: Is an agent an employee or a tool?

Article author, source: 0x9999in1, ME News



• On July 21, 2026, Block officially launched Buzz, an open-source, decentralized, model-agnostic team collaboration platform available on desktop for macOS, Windows, and Linux.
• The most radical aspect of Buzz isn't "like Slack"—it's giving each AI agent a Nostr cryptographic identity, making the agents formal members of the channel, not just plugins.
• The underlying protocol is Nostr, which Dorsey has long bet on; chat, Git hosting, and agent orchestration are all integrated into a single window, directly competing with Slack and GitHub.
• Reference: Slack had approximately 47.2 million daily active users in 2025, holding an 18% market share; Microsoft Teams surpassed 320 million monthly active users. What Buzz aims to seize is not traffic, but the right to define the new frontier of human-machine hybrid work.
• Buzz is the engineering realization of Dorsey’s ideology—open source, autonomy, and deplatforming, consistently upheld.
• But Buzz is not a miracle cure: compliance, SLAs, admin console, and auditing on the enterprise side are all still required—open source does not exempt you from these.
• The real question isn't whether it can replace Slack, but whether it forces Slack, Teams, Notion, and others to answer: Is an agent an employee or a tool?

First, the conclusion: Buzz is not just another Slack clone—it’s Jack Dorsey’s move in the enterprise software market, grounded in a decade-long belief in decentralization.

Pay close attention to the timeline. On July 21, 2026, Block officially announced the launch of Buzz, with immediate full deployment across all three major desktop operating systems. Dorsey’s four words on X were carefully chosen: model-agnostic, decentralized, self-sovereign, fully open-source. Each word strikes at the weakness of today’s SaaS giants.

Why is it called "long-planned"?

The Nostr protocol is not new. Dorsey has been supporting it since 2022, even funding it with Bitcoin. Bluesky is also his creation, though it later split off—sharing the same logic: don’t lock identity, data, or social graphs within any single company. Buzz simply applies this philosophy from the social layer to the workplace collaboration layer.

In other words—Twitter was his past, Bluesky is his present, and Buzz is his future.

And the key keyword along this path has never changed: ownership belongs to the user, not the platform.

Buzz looks like Slack, but its bones aren't Slack's.

On the interface, channels, DMs, and threads are all there. But go one level deeper, and everything changes—it combines Git repositories, issues, and pull requests into a single window, eliminating the need for developers to constantly switch between Slack and GitHub. Any developer who has ever experienced “posting a PR link in Slack, going to GitHub to review it, then returning to Slack to reply” understands this perfectly.

Beneath that lies the real "spark": in Buzz, each AI agent isn't a chatbot—it's a "person." Each has its own account, its own private key, and its own voice. The underlying technology is Nostr—each identity corresponds to a pair of cryptographic keys, inherently portable, auditable, and interoperable across instances.

The New Stack used a precise analogy: Block gave each Agent a "passport." This passport isn't issued by Block—it's held by the Agent themselves. When you take it to other Nostr-compatible tools, it’s still the same "person."

This is completely different from the "AI Copilot" features stuffed into most products today. A Copilot is a button embedded in software; an Agent is a colleague sitting in the meeting room.

How big is the difference?

Copilot serves users, while Agent participates in workflows; Copilot responds passively, while Agent proactively takes on tasks; Copilot operates via API permissions, while Agent operates via identity permissions. The organizational implications behind this are far greater than the technical ones.

To understand Buzz’s ambitions, first consider the size of the market it aims to disrupt.

Slack had approximately 47.2 million daily active users and 79 million monthly active users in 2025, serving over 750,000 organizations, including 77 of the Fortune 100. Salesforce acquired it for $27.7 billion in July 2021, and Slack-related revenue is projected to approach $3 billion in fiscal year 2026. In the global team collaboration software market, Slack holds a steady share of approximately 18%.

Facing it is a larger behemoth: Microsoft Teams, with over 320 million monthly active users. The reason is straightforward—it’s bundled with Microsoft 365, leaving enterprise customers no choice.

Caught between two giants, Slack isn’t having an easy time. It must prove it’s not just “another more expensive chat tool,” but rather the operating system of the AI era. It has launched Slack AI, integrated with Salesforce Agentforce, and invested heavily in embedding Agents directly into the messaging flow.

The issue is that Slack’s agent is “guesting,” not “onboarded.” It has no independent identity, no transferable permissions, and does not persist across systems.

Buzz rushed in precisely to exploit this gap.

Looking back, Dorsey's actions over these years were actually a cohesive strategy.

After Twitter, he largely stepped away from centralized platforms. He works on payments (Cash App, TBD), protocols (Nostr, Bitcoin), and AI tools (Goose, Block’s earlier open-source AI agent framework). These may seem disparate, but they all point to a unified worldview: software should belong to users, identity should belong to users, and data should belong to users.

Buzz is the first time these three puzzle pieces have appeared together in a single product.

Chat layer, decentralized, via Nostr; Collaboration layer, open-source, via Git; Agent layer, model-agnostic, via Goose-compatible framework.

This is a complete transition from philosophy to engineering. You call it idealistic? Yes. But idealism has its place in the enterprise market—especially now, when data sovereignty, AI supply chain security, and model freedom have become essential requirements for large enterprises.

In 2026, how many CIOs will be willing to bet all their agents on a single proprietary SaaS platform? Fewer and fewer.

Setting aside the technical details, let’s ask a more fundamental question: What does it really mean to give an AI agent an ID card?

This means accountability can be established. What the agent did—signatures, timestamps, and action chains—are all traceable. This is essential for audit compliance.

This means permissions can be granular. Since the Agent has its own account, it can be managed as a subject by IT systems—RBAC, ABAC, and the principle of least privilege can all be applied.

This means the Agent can work across platforms. The data analysis Agent I hired today in Buzz can tomorrow take its identity and find work on other Nostr-compatible tools, breaking through the "lock-in effect" of SaaS.

It also means that the boundaries of responsibility will become very blurred.

Is an agent an employee or a tool? When something goes wrong, whose responsibility is it—the team’s, the model provider’s, or the developer who deployed the agent?

This is no longer a technical issue—it’s a legal, organizational, and even an insurance issue. Gartner has repeatedly highlighted in multiple reports for 2026 that "agent identity governance" will be one of the most likely to spiral out of control enterprise risks over the next two to three years.

Buzz didn't solve the problem—it even made it worse. But it was one of the first products to bring the issue to the table and offer a technical solution.

That's enough.

Speaking of this, let’s take a step back for a moment.

Open source, decentralized, agent-first—these terms sound exciting. But in the enterprise software market, victory has never been about technology alone.

Slack has never sold instant messaging. It sells the admin console, SSO, compliance certifications, SOC2, enterprise-grade SLAs, and a dedicated 400 customer service line accessible to major clients. None of these features have been fully delivered by Buzz yet.

Self-hosting is a double-edged sword. It grants you freedom, but also shifts all operational costs, security responsibilities, and upgrade pressures back onto you. Small and medium-sized teams might be tempted by Slack’s $8 per user per month—but CTOs who have truly weighed the costs of IT team hours and compliance audits often hesitate.

Decentralization is even more so. The maturity of Nostr's relay network still lags behind enterprise-grade message queues by an order of magnitude in terms of stability and observability—this is not something that can be bridged in a day or two.

What about the ecosystem? Slack has thousands of official integrations, from Zendesk to Jira to PagerDuty—all included. Buzz is starting from scratch; even if the open-source community moves quickly, it will still take at least 18 to 24 months to build a substantial third-party ecosystem.

Now consider competitor density. AI-native collaboration tools like Centaur are already in the space, and Notion, Linear, and Coda are all evolving toward an agent-native direction. Buzz isn’t the only one seeing the window for human-machine hybrid work—它’s just the most aggressive in its approach.

So, if you ask me whether Buzz will take 10% of Slack’s market share within a year, my answer is: no.

If you ask me whether Buzz will permanently change the answer to the question, "How should agents be integrated into workflows?" within three years, my answer is: very likely.

Slack is defending the Fortune 500, while Buzz is harvesting another group.

The first group consists of development teams that heavily use GitHub. They deeply understand the pain of switching back and forth between a chat window and a Git window. Buzz solves it in one step.

The second category consists of tech companies that have already deployed numerous AI agents. The current reality is that many teams have assembled internal agents using tools like LangChain, AutoGen, and CrewAI, but these agents are scattered across CLI, Notion, Slack, and Discord, making it impossible to coordinate or audit them centrally. Buzz provides a "central station" to solve this.

The third category consists of organizations that are particularly sensitive to data sovereignty and vendor lock-in—European enterprises, public sector entities, Web3-native teams, open-source communities, and some fintech companies. These groups have always been peripheral users of Slack and are natural customers for self-hosted solutions.

The fourth group consists of Dorsey’s followers. Don’t underestimate them. The Nostr ecosystem has already attracted a community of理念-driven developers, and Bluesky has as well. Dorsey’s influence can still rapidly mobilize a core group of seed users within the open-source world.

Together, these four categories aren’t enough to build the next Salesforce, but they’re enough to make Buzz an unavoidable name within 18 months.

The true meaning of Buzz lies not in Buzz itself, but in the direction it represents.

What has been the SaaS logic over the past two decades? Package the features, lock in the data, and turn user identities into your paid seats. Slack, Notion, Figma, Salesforce—none are exceptions.

The explosion of AI agents is tearing a hole in this logic.

Because agents are inherently cross-system. Today they read your Notion, tomorrow they send Slack messages, the day after that they create Jira tickets, and the day after that they call the GitHub API. They don’t belong to any single SaaS—they belong to the workflow itself.

And the carrier of the workflow is identity.

Whoever controls the identity of the Agent controls the gateway to the next generation of collaboration.

Slack wants to hijack this entry point with "Agentforce + Slack messaging flows." Microsoft wants to hijack this entry point with "Copilot + Entra ID." Google wants to hijack this entry point with "Gemini + Workspace."

Dorsey said: No, identity should belong to the agent itself, to the user themselves, and to the protocol itself.

This is a battle over what the next generation of enterprise software should look like. Buzz is the sharpest bullet in this war.

It may not hit the bullseye, but it forces every giant to take another look at their own muzzle.

Dorsey was never the type to build a product worth $10 billion. He’s the type who introduces an idea into the market and forces everyone else to confront it.

Twitter started this way. Cash App started this way. Nostr started this way. Buzz did too.

He doesn't need Buzz to outperform Slack. He just needs Buzz to make every SaaS CEO wake up in the middle of the night wondering: If my users truly want self-sovereign collaboration tomorrow, can I still keep up?

Some might say this is just another idealistic gesture. But looking back over the past decade, everything Dorsey has seriously bet on—payments, protocols, social media, Bitcoin—has never truly been ignored by the market.

In terms of enterprise collaboration, it’s likely unavoidable going forward.

On its first day, no one knew what Buzz would look like three years later. But one thing can be written down now:

Once you send your ID to an AI agent, there's no going back.

The rest is just a matter of time.

Reference materials

1. TechCrunch, *Jack Dorsey is taking on Slack with Buzz, a group chat platform for teams and their AI agents*, 2026-07-21.
2. The New Stack, *Block built a Slack for AI agents—and gave each one its own passport*, 2026-07-22.
3. SiliconANGLE, *Block launches Buzz, an open-source workspace for humans and AI agents*, 2026-07-21.
4. Crypto Briefing, *Jack Dorsey's Block launches Buzz group chat platform to challenge Slack and GitHub*, 2026-07-21.
5. DemandSage, *Slack Statistics 2026 [Revenue & Market Share Data]*, 2026.
6. Business of Apps, *Slack Revenue and Usage Statistics (2026)*, 2026.
7. TechRT, *Slack Statistics 2026: Powerful Usage Insights* (including Microsoft Teams' 320 million MAU data), 2026.
8. Quantumrun, *Who Owns Slack?* (Salesforce's $27.7 billion acquisition and Slack's $3 billion revenue expectation for fiscal year 2026), 2026.