Meta unveiled a new enterprise business built around its AI models and agents on Monday, tapping MongoDB CEO CJ Desai to run the division as Chief Enterprise Platform Officer. The initiative, named Meta Enterprise Platform, will package technology Meta developed for its consumer applications and advertisers into products that companies and developers can use within their own operations. Mark Zuckerberg characterized the effort as the "next major pillar" of the company's business, placing enterprise AI alongside Meta's advertising and consumer app divisions.
Desai departed MongoDB with immediate effect, less than twelve months after taking the CEO role in November 2025, with former CEO Dev Ittycheria stepping in as interim president and CEO. For developers, Meta plans to offer its "full technology stack," beginning with the Muse agent, Meta Business Agent, Muse API, and Muse Code. The company introduced Muse earlier this month as a consumer-facing personal AI agent, while Meta Business Agent—launched in June—manages customer interactions for companies across Meta's platforms. Muse API and Muse Code target developers directly, putting Meta in tighter competition with OpenAI, Anthropic, and Google for engineering teams building agents and coding workflows. Though developers can already access both tools—Muse Code has been in beta since August, and Meta started charging for its Muse Spark model through its API in July at $1.25 per million input tokens and $4.25 per million output tokens—the company didn't release enterprise pricing, general availability dates, or service terms on Monday.
According to the report, Llama, the model family Meta promoted for years as its open-weights option for teams wanting to run and fine-tune models on their own infrastructure, doesn't appear anywhere in Meta's description of the new enterprise stack. Meta hasn't clarified whether Llama will be part of the Enterprise Platform, though it has released open weights for its smaller Muse Glimmer model and promised an open-weights version of Muse Spark. In a statement, Desai said Meta Enterprise Platform will focus on converting Meta's AI stack into products and services that companies can deploy inside their own businesses, with security and privacy built into Meta's enterprise products "from the outset." However, Monday's announcement didn't address data retention, training on customer data, tenant isolation, identity controls, or compliance certifications.
The move puts Desai's experience building AI data infrastructure at the center of Meta's enterprise push. At MongoDB, he led the addition of persistent agent memory, automated embeddings, and other AI capabilities to the data platform in May, arguing that the model represents only part of the challenge and that bringing agents into production depends heavily on the data layer behind them. Meta faces a similar problem: agents operating inside a business need persistent context, access to constantly changing data, and a way to act inside the applications employees already use. Desai's earlier roles included leading product and engineering at Cloudflare and spending nearly eight years at ServiceNow, eventually becoming president and COO. Meta already has relationships with many of the businesses it wants to reach through the billions of people who use its products and the hundreds of millions of businesses on its platforms, many of which are small and midsize companies that already use Facebook, Instagram, and WhatsApp for marketing and customer service.
The report notes that Meta Enterprise Platform is "more strategy than product" for now, even though some of its components are already in developers' hands. The announcement leaves one big question unanswered: whether Meta's enterprise AI future still includes Llama, or whether Muse signals a shift toward a more controlled platform where developers access Meta's latest agent technology through Meta itself, even as Meta promises open weights for Muse Spark. Winning over developers at larger companies will be harder, since their engineering teams have spent the past several years building around models and platforms from OpenAI, Anthropic, Google, and others, and Meta needs to offer something that gives teams more than reach before they add another platform to their stack or move over completely. The hire signals Meta's recognition that selling enterprise AI requires more than powerful models—it demands the infrastructure expertise to make those models work inside existing business systems. Whether developers will trust Meta with that role remains the platform's defining test.