What Is Buzz? Jack Dorsey's Open-Source, Agent-Native Slack Rival
Buzz is an open-source Slack alternative from Block where AI agents work as first-class team members. Here's how it works and why it's different.
What is Buzz?
Buzz is an open-source team chat application built by Block, the company led by Jack Dorsey (also known for Twitter and Square). It looks and functions like Slack, with channels, threads, replies, and emoji reactions, but it’s built around a different premise: AI agents are treated as first-class users, not bolted-on bots. Buzz is self-hostable, privacy-focused, and gained roughly 27,000 GitHub stars within about a week of launching, signaling strong early interest from developers.
TL;DR
- Buzz is an open-source Slack alternative from Block that lets teams self-host their own chat infrastructure instead of depending on a SaaS vendor.
- Its core idea is agent-native design: AI agents join channels and do work alongside humans as equal participants, not as add-on integrations.
- Every action in Buzz, whether a message, reaction, workflow step, approval, or agent event, is logged as a signed event using Nostr relay infrastructure, giving one unified audit trail for people and processes.
- Buzz supports workflow and automation building directly inside the app, positioning itself as more than a messaging tool and closer to a work hub.
- It’s model-agnostic, meaning teams can plug in open-source models, closed models, hosted APIs, or local models rather than being locked into one provider.
- Because it’s backed by Block, a company with real engineering resources, Buzz is likely to see continued development rather than being an abandoned side project.
- The tool is still new, so early adopters should expect rough edges typical of a fast-moving open-source release.
Remy is new. The platform isn't.
Remy is the latest expression of years of platform work. Not a hastily wrapped LLM.
How does Buzz treat AI agents differently than Slack does?
Slack supports bots and app integrations, but they exist in a separate tier from human users. They post messages, respond to slash commands, or run in the background, but they’re clearly guests in a system designed for people.
Buzz flips that model. Agents in Buzz occupy the same channels as humans and are represented with the same kind of identity and permission structure as a person. A team member and an AI agent can both work a task in the same thread, react to the same messages, and show up in the same audit log with equivalent standing. The distinction between “a person did this” and “an agent did this” becomes a metadata detail rather than an architectural divide.
This matters because as more teams start using autonomous coding agents, research agents, and workflow bots day to day, chat tools built around a human-only mental model start to feel like they’re working against the grain. Buzz is an attempt to build the chat layer assuming agents are already part of the team.
Why does Buzz log everything as a signed event?
One of the more distinctive engineering choices in Buzz is its use of Nostr relay infrastructure for its underlying event log. Nostr is a decentralized protocol originally built for social networking, based on simple, signed, cryptographically verifiable events.
In Buzz, every message, reaction, workflow step, review, approval, and agent action becomes a signed event in that same log. The system doesn’t distinguish structurally between a human sending a message and an agent completing a workflow step: both are just events with the same shape, the same identity model, and the same audit trail.
This design has a practical payoff for teams running agents in production. If an agent takes an action inside Buzz, that action isn’t a black box buried in some separate system log. It’s recorded the same way a human message would be, with a verifiable signature tied to an identity. For teams worried about accountability as agents get more autonomy, that unified, tamper-evident record is a meaningful feature, not just a technical curiosity.
Is Buzz just a chat app, or something more?
Buzz is designed to go beyond messaging. It embeds workflow and automation building directly into the platform, so instead of treating chat as the front end and automations as a separate system (think Zapier or a workflow engine bolted on the side), Buzz wants to be the place where both live together.
The pitch is that Buzz becomes a central hub: conversations happen there, decisions happen there, and the automations and agent workflows that act on those decisions run there too. That’s a more ambitious scope than a typical Slack clone, and it’s consistent with where a lot of AI infrastructure is heading: fewer disconnected tools, more single surfaces where humans and agents collaborate and where the system of record for both is the same.
Can you self-host Buzz, and does it lock you into one AI model?
Yes. Buzz is open source and built to be self-hosted, which appeals directly to teams and organizations that care about data privacy and control over their infrastructure. Rather than routing every message through a third-party SaaS platform, organizations can run Buzz on their own servers and keep full ownership of their data and conversation history.
On the model side, Buzz is model-agnostic. Teams aren’t locked into a single AI provider. You can connect open-source models, hosted commercial models, or fully local models depending on your needs, budget, or compliance requirements. That flexibility matters for teams that want agent behavior powered by whichever model fits a given task, rather than being stuck with one vendor’s stack because the chat tool only supports it.
Is Buzz worth trying right now?
Buzz is genuinely new, and being new comes with the usual caveats: expect a smaller plugin ecosystem, less battle-tested infrastructure, and likely a faster pace of breaking changes than an established product like Slack. It also doesn’t yet have the years of enterprise hardening, compliance certifications, or third-party integration marketplace that Slack has built up.
That said, the fast growth in stars right after launch suggests real developer interest, and Block has the engineering resources and reputation to keep pushing the project forward rather than letting it stall. For teams already experimenting with agent-driven workflows and looking for a chat platform that treats agents as native participants rather than bolted-on bots, Buzz is worth testing in a low-stakes setting to see whether the agent-native model fits how the team actually works.
Frequently Asked Questions
What company makes Buzz?
Buzz is made by Block, the company led by Jack Dorsey, who also co-founded Twitter and Square.
Is Buzz free to use?
Yes. Buzz is open source, which means it’s free to download, inspect, modify, and self-host on your own infrastructure.
How is Buzz different from Slack?
The core difference is that Buzz treats AI agents as first-class users in the same channels as humans, with equal standing in permissions, identity, and the audit log. It also builds workflow automation directly into the platform and uses a signed event log (via Nostr relay infrastructure) to track every action, human or agent, in one unified record.
Do I need to use a specific AI model with Buzz?
No. Buzz is designed to be model-agnostic, so you can connect open-source models, hosted commercial models, or local models depending on your setup.
Is Buzz ready for production use?
It’s a newly launched project, so treat it as early-stage. It has drawn fast attention on GitHub and has backing from Block, but teams should expect the kind of rough edges typical of a young open-source tool before relying on it for critical workflows.
