Jack Dorsey launches Buzz: Slack + Git + AI agents in one place - a coherent bet, or a full-stack fantasy?

In plain terms - Jack Dorsey's Block launched Buzz on 21 July 2026: a single platform combining team chat, AI agents, and Git hosting. The pitch: escape Slack + GitHub + LLM-tool fragmentation, use Buzz for all three.

Contexte

The trend Buzz is riding is real: teams do juggle Slack, GitHub, a Copilot-adjacent tool, and increasingly an agent runtime. Every seam between them is friction - permissions, notifications, context that doesn't travel. The counter-trend is equally real: nobody rips out GitHub. Or Slack. Or their existing agent stack.

Under the hood

• Chat: team-messaging surface, feature parity with Slack basics at launch (channels, threads, DMs).
• Git hosting: full Git server + PR/review UI. Comparable to a basic GitLab install.
• AI agents: agents runnable inside channels, with awareness of the Git context (repo, PRs, issues).
• Integration point: the three surfaces share a single permissions/identity model, which is Buzz's argument for being sticky.

Analyse

The bet is architectural, not just product. Slack + GitHub + agent tool is three separately-permissioned systems, and the agent is the piece that most needs unified context (it wants your Slack thread AND your PR AND your issue tracker). Buzz collapsing those into one identity/permission model is a genuinely thoughtful design move. It's how "agent-first team platform" would have been built if the stack had been rebuilt from scratch in 2026.

But: the switching cost of GitHub is not the product - it's the ten years of PR history, CI wiring, and dependency graph living inside org accounts. Buzz can't replicate that overnight, no matter how good the chat surface is.

Two paths for Buzz's realistic wedge:

1. Greenfield teams / new startups - where GitHub inertia is minimal and "agent-native from day one" is a real pitch.
2. Air-gapped/self-hosted - where organisations already run their own Git and Slack alternatives, and consolidating on Buzz is fewer systems, not more.

For an incumbent team, the interesting question is whether the agent surface alone (chat + repo-aware AI) is enough to justify running Buzz alongside Slack/GitHub for the agent work only. That's how Zoom entered against Skype - a wedge, not a replacement.

Scénarios

• Base case (55%): Buzz finds a wedge in agent-first team platform for new/small teams; doesn't dislodge Slack or GitHub in incumbents.
• Fast take (20%): Block's brand + Dorsey's distribution pulls enterprise pilots; carve-out becomes serious in 12 months.
• Fade (25%): too many surfaces to build simultaneously; execution slippage on any one kills the differentiator.

Implications

For an engineering leader: don't switch - evaluate. Buzz is worth a pilot on a single team's agent workflow, not a wholesale migration. For a Slack/GitHub incumbent: the "AI-native rewrite" is now a shipping product, not a threat in slides.