The irony is perfect: Meta just paid Salesforce's Slack a premium to become the testing ground for the AI agents that might eventually kill Salesforce's core business.
The Summary
- Meta is migrating from Google Chat to Slack for internal communications, explicitly because Slack is "the strongest platform available today for agents," according to a memo from AI chief Alexandr Wang
- The move validates Slack's bet on becoming agent infrastructure rather than just a communication tool, while highlighting that the platform wars are now being fought on agent compatibility
- Meta's 70,000+ employees will become the largest single testbed for workplace AI agents, and what works there will define expectations everywhere else
The Signal
Meta didn't switch chat platforms because Slack has better emoji or thread organization. They switched because Slack has become the de facto operating system for workplace AI agents. Wang's memo is blunt about this: "rich conversational interface for agents, mature developer tooling and strong 3P integrations." Translation: when you want to deploy dozens of AI agents across your organization, Slack is the only platform with the plumbing already built.
This matters because Meta is essentially saying the future workplace isn't human-to-human chat that occasionally uses AI features. It's human-to-agent and agent-to-agent communication that happens to include humans. The company that built the social graph for 3 billion people just declared that their internal collaboration graph needs to be agent-native.
"While not all of us are building agents today, everyone at the company will benefit from a robust and useful agent ecosystem."
The timing reveals where we are in the agent adoption curve. Salesforce bought Slack for $27.7 billion in 2021, back when it was still primarily a human communication tool. Five years later, that acquisition looks either brilliant or desperate depending on how you read it. Brilliant because Slack positioned itself as agent infrastructure before "agent infrastructure" was a category. Desperate because Salesforce's stock has fallen over the past year on fears that AI agents will replace the CRM and productivity software that built the company.
Meta's migration is Salesforce's proof point that the second path is the right bet. If AI agents need somewhere to live and operate, and that place is Slack, then Slack becomes more valuable even as standalone software becomes less valuable. The company's stock surged 23% in a single day last week, and this Meta deal is exactly why.
Key strategic shifts this reveals:
- Platform value is migrating from features to agent compatibility
- The "best" workplace tool is now the one that works best for non-human workers
- Third-party integrations matter more than native features when agents are the users
What Google Chat missed is instructive. Google has deeper AI capabilities than Salesforce. They have Gemini, they have infrastructure, they have the talent. But they didn't build Google Chat as agent substrate. They built it as a cheaper alternative to Slack for human teams. Meta's defection suggests that strategy is already obsolete.
The Implication
Watch which platforms your company's AI agents can actually use, not which ones have the shiniest AI features bolted on. Meta just told you that agent infrastructure matters more than AI capability. The companies that win the next five years will be the ones that built their products assuming agents would be the primary users, with humans as the occasional supervisors.
If you're building workplace tools and you're still optimizing for human UX first, you're building for a market that's shrinking. The race is now about who can be the runtime environment for the agent economy.