Meta announced the Meta Enterprise Platform on Saturday, September 28. MongoDB CEO CJ Desai joined Meta as chief enterprise platform officer, reporting directly to Mark Zuckerberg. The platform will initially prioritize tools including Muse agent, business agent, and a coding tool. Zuckerberg said the platform uses advanced models, leading agents, large-scale infrastructure, and years of working closely with many businesses. MongoDB stock fell 25% on the news. Desai had been MongoDB CEO for less than a year. The announcement came three weeks after Meta launched Muse, the personal AI assistant that can book travel and send email. Meta now sells enterprise AI. The API I call from production does not know that yet.
The dog-training site at valhalla-k9.com does not call Meta's APIs. The ERP at coenconstruction.com does not call Meta's APIs. The permit tool does not call Meta's APIs. I call OpenAI for invoice extraction, Anthropic for blueprint parsing, Google for zoning summaries. If I called a Meta endpoint today, I would get a consumer product packaged as an API. The endpoint does not surface an enterprise tier. The response does not include an SLA. The rate limit does not distinguish between a hobbyist running a Discord bot and a contractor parsing twelve hundred invoices per month. Desai was in his role at MongoDB for less than a year. He spent nearly eight years at ServiceNow and a year at Cloudflare before that. He knows how to sell software to businesses. He does not control the API I call today.
The hire signals intent but not availability
Desai received an offer directly from Zuckerberg that he could not refuse; the package from Meta was enticing enough for Desai to give up his pay package and severance at MongoDB. Meta hired an enterprise operator to build an enterprise business. That business does not exist in the API yet. When OpenAI launched ChatGPT Enterprise in August 2023, the product came with guaranteed uptime, dedicated capacity, and priority support. The API included new endpoints for enterprise customers. The response metadata included tenancy information. The rate limits were higher. The changelog documented the enterprise features in the same place it documented the consumer features. The business-systems operator could read the diff and decide whether to upgrade. The price drops, the latency improves, the context window doubles, but the migration path stays the same.
Meta announced an enterprise platform on September 28. The announcement does not include API documentation. The announcement does not include pricing tiers. The announcement does not include availability dates. The business launched with an executive hire and a press release. The API will follow later, or it will not. Desai said AI will fundamentally redefine how organizations of all sizes innovate, grow, serve customers, and run business operations. The construction ERP redefines how subcontractors submit invoices and how project managers approve pay applications. The permit tool redefines how installers pull zoning summaries and setback requirements. Those redefinitions depend on API calls that return JSON objects in under two seconds with a known error budget. The enterprise platform is an organizational initiative. The API is a software artifact. The two move on different timelines.
The platform targets businesses Meta already serves
Meta already serves hundreds of millions of businesses through Facebook and Instagram advertising. Small businesses are fertile ground for Meta due to its deep ties to them through Facebook and Instagram. The businesses that buy ads are not the businesses that call APIs. The ad buyer wants a dashboard that shows impressions, conversions, and cost per acquisition. The integration engineer wants an endpoint that accepts a schema, returns structured data, and logs errors to a webhook. The consumer product and the business product live in different repositories with different teams and different contracts. Meta sells ads to small businesses. Meta will sell agents to small businesses. The agent may call the same model the API exposes, or it may not. The enterprise platform will serve businesses that use Facebook and Instagram. The API will serve developers who build software. Those are not the same customer.
The CEO left after ten months. The stock dropped 25%. The platform is enterprise. The API call still hits the consumer tier.
Desai had been in the CEO role for less than a year, having taken over from Ittycheria in November 2025. He left MongoDB for Meta because the offer was large and the opportunity was rare. Meta is betting that enterprise customers will pay for AI agents the way they pay for advertising. The bet does not change the fact that the Muse API—if I called it today—would return a consumer response with consumer limits and consumer error handling. The platform commitment is strategic and the deployment is tactical. The platform runs on a timeline measured in quarters and executive hires. The API runs on a timeline measured in milliseconds and HTTP status codes. I will know the enterprise business is real when the API documentation includes an enterprise tier, the pricing page shows business pricing, and the response headers include tenancy metadata. Until then, the platform is a press release and the API is a consumer product.