The first major chat platform built for humans and AI agents working together just launched, and it's open source by default.
The Summary
- Block launched Buzz, an open source group chat platform designed for humans and AI agents to collaborate across messages, code, and workflows
- Buzz positions itself as a decentralized alternative to Slack, blending chat with code collaboration traditionally split between Slack and GitHub
- This is the first mainstream enterprise tool purpose-built for human-agent teams, not just human-to-human communication with AI features bolted on
The Signal
Block's Buzz isn't just another Slack clone with a chatbot plugin. It's a platform where AI agents are first-class citizens alongside humans, collaborating across messages, code repositories, and workflow automation in a single interface. The distinction matters. Slack treats AI like a fancy add-on. Buzz treats agents like teammates who need their own permissions, workflows, and collaboration patterns.
The open source foundation signals something bigger than product strategy. Dorsey has spent years pushing for protocols over platforms, and Buzz extends that philosophy to workplace collaboration. By making the code public, Block is betting that developers will fork, extend, and build agent-native features faster than any closed platform could ship them.
"The first mainstream enterprise tool purpose-built for human-agent teams, not just human-to-human communication with AI features bolted on."
The timing aligns with a shift already underway. Companies are moving from "we have an AI strategy" to "our agents need infrastructure." Chat platforms built for human conversations break when you add agents that process thousands of messages per hour, trigger workflows across multiple systems, and need audit trails that humans don't. Buzz acknowledges this reality in its architecture.
What makes this different from Microsoft's Copilot-everywhere approach:
- Open source vs. proprietary lock-in
- Agents as peers vs. assistants
- Code and chat unified vs. separate tools
The decentralized positioning matters less than the practical architecture. Most teams won't care about decentralization until they need it. They'll care that their deployment agent can spin up environments directly from chat, or that their monitoring agent can post alerts with full context and proposed fixes. That's the wedge.
The Implication
Watch how quickly agent-first features ship on Buzz versus Slack's roadmap. If developers start building agents that only work well on Buzz, the network effects flip fast. Slack won on integrations. Buzz could win on agents that actually integrate themselves.
For anyone building AI agents, this is your signal to design for multi-agent environments from the start. The workspace of 2026 isn't "humans with AI help." It's humans and agents working parallel threads, with tools that respect both workflows.