Meta puts Muse models, agents and coding tools under a new enterprise platform
Mark Zuckerberg named former MongoDB CEO CJ Desai to lead the effort, which brings Meta's AI models, business agents and developer tools to businesses.
By Ryan Merket · Published
Primary source: X
Why it matters
Meta is organizing models, customer-facing agents and developer tools into a business-facing initiative under a veteran enterprise operator. The announcement names no enterprise revenue or customer commitments, leaving adoption to prove whether this becomes a meaningful business beyond Meta's ad engine.
Mark Zuckerberg said Meta is starting a new enterprise platform on September 28th, putting its Muse models, AI agents and developer tools into a business-facing push led by former MongoDB chief executive Chirantan "CJ" Desai. Zuckerberg described the effort as the "next major pillar" of Meta's business and said Desai will report directly to him.
The move pulls together products Meta had already begun introducing separately. In June, Meta announced its Business Agent Platform, which lets companies build and deploy customer-facing agents on WhatsApp, Messenger and Instagram. The company said at the time that more than one million businesses were already using a Meta Business Agent on WhatsApp and Messenger. Meta's September 8th Muse launch introduced a personal agent for consumers. Zuckerberg's new platform pitch extends the same family of technology to business buyers and developers.
The product list in Zuckerberg's four-post thread includes the Muse agent, Meta Business Agent, Muse API and Muse Code. Meta's Muse Code documentation describes a terminal and continuous-integration coding agent that can plan tasks, edit files and run commands, with user approvals and an operating-system sandbox. Meta's developer overview says developers can also call Muse models directly through an API or connect them to coding agents using OpenAI- and Anthropic-compatible interfaces.
Those tools give the announcement more substance than a new division name: Meta has public-facing products and documentation for several parts of the proposed stack. But the thread does not set out enterprise pricing, customer commitments or a timetable for delivering the platform as a unified offering. Zuckerberg says Meta intends to bring its "full technology stack" to businesses; the post does not describe what customers will buy first or how the products will be packaged.
Meta has begun publishing usage-based Model API pricing. Its standard Muse Spark tier is listed at $1.25 per million input tokens and $4.25 per million output tokens; a lower-priced Contributor tier permits Meta to use prompts and completions to train future models. That is a developer API price, not an announced price for the broader Enterprise Platform. The distinction matters for buyers evaluating a model endpoint versus a managed agent system with business controls and support.
Meta's existing business already supplies a large distribution channel. The company reported $200.97 billion in 2025 revenue, of which $196.18 billion came from advertising, according to its full-year results. Zuckerberg's thread says Meta helps hundreds of millions of businesses reach customers. The enterprise effort would give Meta another route to sell AI services to companies that already use its platforms, while placing its agents and models in competition with offerings from established cloud, software and AI vendors. Whether that turns into material revenue will depend on business adoption and spending, neither of which the announcement quantified.
Desai brings experience running large software organizations, rather than a research-lab profile. MongoDB announced him as its president and CEO in November 2025, effective November 10th. Before MongoDB, he led product and engineering at Cloudflare and served as president and chief operating officer at ServiceNow; MongoDB's leadership biography also lists earlier roles at EMC, Symantec and Oracle. The hire gives Zuckerberg an executive with enterprise operating and go-to-market experience to lead a product set that now spans infrastructure, agents and developer access.
That appointment is the clearest indication that Meta wants enterprise AI treated as a coordinated commercial business. The underlying pieces have been appearing in consumer and business channels; now Zuckerberg is putting one executive over their enterprise direction. The immediate test is whether Meta can convert that collection of models and agents into a coherent offer companies will deploy at scale.