On September 28, 2026, Mark Zuckerberg announced Meta Enterprise Platform, a unit that will turn Meta's AI stack into products businesses and developers can deploy. The initial lineup is the Muse personal agent, Meta Business Agent for customer conversations and tasks, the Muse API (self-serve access to the Muse Spark reasoning model through the Meta Model API), and Muse Code, a terminal coding agent that orchestrates multi-agent workflows and is now out of beta. The platform is led by Chirantan "CJ" Desai, Meta's new chief enterprise platform officer, reporting directly to Zuckerberg. He was most recently CEO of MongoDB, and before that held product and engineering leadership at Cloudflare and spent nearly eight years as president and COO of ServiceNow.
What Meta has not published matters as much. The announcement names no enterprise pricing, no launch date for the enterprise products, no launch customers and no regions. The per-token Muse Spark API pricing is already public separately: Runtime Wire reports $1.25 per million input tokens and $4.25 per million output. So developers can build today, but an enterprise buyer can't yet procure a packaged product.
Why it matters: Meta reported $200.97 billion in 2025 revenue, $196.18 billion of it from advertising. This is its first serious attempt at a large AI revenue line that isn't ads, and hiring an enterprise-software operator to run it signals a direct sales motion, not just an API. Meta's distribution edge is real. It said in June that more than a million businesses already use a Business Agent on WhatsApp and Messenger, which gives it an installed base most enterprise AI entrants would pay for.
The open questions are the ones enterprise procurement will ask first. How is customer data separated from the advertising business? Zuckerberg's post says "security and privacy are built into Meta's enterprise products from the outset", and buyers will want that in contract terms and audit rights, not just a blog post. Can Meta's support and compliance organisation match vendors that have sold to regulated industries for decades? For developers, the practical effect is more price pressure on frontier-class APIs. Watch for the first named enterprise customers and for published data-handling terms.
Meta is building its first major non-advertising AI business on top of a million-business WhatsApp/Messenger base, but with no enterprise pricing, customers or data-separation terms published yet, it is a developer API today and an enterprise product later.