More or Less · Topics · Episode #165

OpenClaw 2.0: Multiplayer Coding — and Is Slack Really an Agent Platform?

Clip from the full recording, 27:42–31:16 · download mp4 · watch at this point on YouTube

Dave ships OpenClaw 2.0 after two months: a multiplayer agentic-engineering interface he claims now beats Codex and Claude Code's web UI. Then the table splits over Slack — Meta is adopting it for agents, Dave says it's objectively the worst agent platform, and Sam declares MCP isn't a thing.

Dave’s product spotlight: OpenClaw 2.0 shipped Sunday night after a two-month push, and the headline is multiplayer. His team has been building OpenClaw with OpenClaw, and the collaboration was previously duct tape — everyone’s individual instances, GitHub, Discord, Slack. The release is effectively a multiplayer Codex: an entire engineering team leveraging agents and tokens in one interface. Dave claims the web interface is now genuinely better than Codex and Claude Code’s web UI, and Brit points out the domestic upside: Sam and Jess can now build apps together.

Then the fight. Jessica notes Meta is reportedly moving to Slack because it’s better for agents, and Dave — a Slack seed investor — erupts: “objectively, this is not true… you put two agents in the same channel on Slack, they cannot communicate without a bunch of weird software in the background.” Brit takes the other side: Slack has the MCP, the integrations, and most importantly enterprise trust — companies dabbling in agents will start where they already live, which makes Slack the leader today whether or not it’s the best. Sam’s contribution is a grenade: “No one cares about MCP. MCP isn’t a thing.” Jessica, mildly: she uses Slack’s MCP. The segment is a clean snapshot of the agent-platform land grab: the purpose-built tool versus the incumbent with distribution.

Key points

  • OpenClaw 2.0 shipped Sunday after two months: multiplayer agentic engineering — a full team sharing agents and tokens in one interface.
  • Dave's claim: the OpenClaw web interface now beats Codex and Claude Code's web UI.
  • The team builds OpenClaw with OpenClaw — the release replaces a duct-tape stack of GitHub, Discord, Slack, and individual instances.
  • Meta is reportedly adopting Slack as its agent-collaboration platform — which Dave, a Slack seed investor, calls objectively wrong: two agents in a channel can't talk without background glue.
  • Brit's counter: trust and distribution beat architecture — enterprises will start their agent journey inside the tool they already trust, and today that's Slack.
  • Sam's grenade: "No one cares about MCP. MCP isn't a thing." Jessica dissents — she uses it.

Where they landed

DaveOpenClaw 2.0 is the real agent-collaboration layer; Slack is the worst possible one, Meta's move notwithstanding — and he says that as a Slack seed investor.
BritSlack wins on trust and integrations for now; big cos will dabble where they already live. 'Not for long' is Dave's problem to prove.
SamMCP is a non-event, and the multiplayer future matters more than which chat app hosts the bots.
JessUses Slack's MCP, notes Meta's endorsement — maybe it's just better than whatever else exists right now.

Quotes

“You put two agents in the same channel on Slack, they cannot communicate without a bunch of weird software in the background.”— Dave
“I'm now absolutely in love with the OpenClaw web interface. I think it's actually better than Codex now.”— Dave
“No one cares about MCP. MCP isn't a thing.”— Sam

Suggested tweets

OpenClaw 2.0 is out: multiplayer agentic engineering — a whole team sharing agents in one interface. Dave Morin claims it now beats Codex

Tweet this →

Meta says Slack is its agent platform. Dave Morin — a Slack seed investor — says that's objectively wrong: two agents in a channel can't even talk

Tweet this →

“No one cares about MCP. MCP isn't a thing.” — Sam Lessin, starting fights in one sentence

Tweet this →

All topics · Full episode #165 + transcript · Subscribe on YouTube · Spotify