Meta Launches an Enterprise AI Unit, Betting Muse Can Win Business Buyers

Meta launched Meta Enterprise Platform on Monday, September 28, a new business unit that will sell its AI models and agents to companies, and hired MongoDB's chief executive to run it. Mark Zuckerberg called the unit "the next major pillar of our business" in Meta's announcement. For CIOs and chief AI officers, the move adds a well-funded new vendor to an already crowded market. It also raises a fair question: can a company built on consumer apps and advertising earn the trust that enterprise AI buying now requires?

What is Meta Enterprise Platform?

According to Meta, the unit will start by bringing Meta's full technology stack to businesses and developers. The named offerings are the Muse agent, Meta Business Agent, the Muse API and Muse Code. Meta did not give pricing, plans or an availability schedule.

VKTR described what each piece does. Muse is a personal agent that runs on a dedicated cloud virtual machine and can browse, fill out forms and complete tasks. Meta Business Agent is a customer-facing agent introduced in June. The Muse API gives developers access to Muse models, and Muse Code is a terminal-based coding agent that has been in beta since August. VKTR also reported that a Muse Confidential VM, which encrypts the whole environment with a key only the user holds, is planned for later this year.

A day later, on September 29, Meta added Muse for Small Business, which connects to tools such as Stripe, QuickBooks, Shopify and Slack and, Meta says, takes no action that publishes, sends or spends money without the owner's approval. Most features are free.

Who is CJ Desai, and why does the hire matter?

Chirantan "CJ" Desai becomes chief enterprise platform officer and reports directly to Zuckerberg. Before MongoDB, he led product and engineering at Cloudflare and spent nearly eight years at ServiceNow, including as president and COO, Meta said. That is a resume built on selling to large IT organizations, experience Meta's advertising business lacks.

Investors in MongoDB took the news hard. The stock fell about 18% in morning trading on Monday, Reuters reported, and The Motley Fool put the intraday low at nearly 27% before a partial recovery. Desai had led MongoDB for less than a year. Former CEO Dev Ittycheria is returning as interim chief executive. Piper Sandler analysts called the departure disappointing but described the sell-off as an overreaction, according to Reuters.

Desai said the unit will focus on services companies can deploy for their own businesses, and that security and privacy are built into Meta's enterprise products from the start, according to Meta's announcement.

What is the case for Meta as an enterprise AI vendor?

Meta's argument rests on scale and momentum. Muse, launched earlier in September, climbed app store rankings quickly, and some analysts have called it potentially the biggest US app launch since ChatGPT, Reuters reported. Desai pointed to Meta's models, infrastructure, agents and its relationships with millions of advertisers and hundreds of millions of businesses. Many companies already run customer conversations on WhatsApp, Messenger and Instagram. In June, Meta said more than a million businesses used earlier versions of its chatbots on WhatsApp and Messenger, according to a Reuters report published by TechCentral, and it set up an enterprise solutions team that embeds engineers with customers.

There is also a design point in Meta's favor. VKTR noted that the safeguards in the consumer version of Muse, including an isolated cloud computer, a separate agent that approves outbound traffic and audit trails, line up with what enterprise security teams usually ask for. Meta has not said which of them carry over to the enterprise platform.

What are the reasons for caution?

The skeptical view starts with the sale itself. Agents that act inside company systems must pass security reviews, procurement and compliance checks, a very different process from selling ads, VKTR pointed out. Meta also acknowledged gaps in coding and long-horizon agent tasks when it unveiled its Muse Spark model in April, and Muse Code competes directly with established coding agents from OpenAI and Anthropic.

Trust is the second issue. About a week before the launch, Amazon blocked Muse from shopping on its site, saying Meta had not notified it, that Muse does not identify itself when browsing and that it appears to capture and store customer credentials, VKTR reported. Meta said earlier that Muse cannot see users' passwords or payment methods and that shared credentials sit in secure storage. In June, attackers talked a Meta AI support chatbot into handing over access to high-profile Instagram accounts; Meta said at the time that its investigation pointed to a flawed technical check in the underlying support function, Reuters reported.

The open question enterprise buyers will raise first is data. Meta has not yet said in detail whether enterprise data stays out of model training and out of its advertising systems.

What should enterprise leaders do now?

Meta's entry is good news for buyers in one sense: more serious competition tends to bring better pricing and terms. But a new vendor with no published pricing, no enterprise data commitments and an unproven agent stack is a candidate for evaluation, not for a core workflow.

Practitioners at AI Connect events have argued for keeping options open. "You should actually architect for having no idea where things are going to go," Girish Bhat, SVP at Revefi, said during a AI Connect webinar on choosing AI use cases and build-versus-buy decisions. An earlier panel on avoiding vendor lock-in in AI infrastructure recommended open frameworks and testing several vendors in proofs of concept.

Three practical steps follow. Put Meta's written answers on training, ad separation, data residency and retention into the first round of any evaluation. Treat Muse agents like any other machine identity, with scoped access and an audit trail, a principle NexusOne CEO Anu Jain summed up at COLLIDE 2026 this week: "AI needs complete context and hard boundaries," as described in our coverage of his talk on data and AI control planes. And if your company already uses Meta's business messaging, pilot Meta Business Agent there first, where the data relationship already exists and the value can be measured.

Watch for Meta's enterprise pricing, its first named customers and partners, and the data terms it is willing to put in contracts.

Frequently asked questions

What is Meta Enterprise Platform?

It is a Meta business unit launched September 28, 2026, to sell Meta's AI stack to companies, starting with the Muse agent, Meta Business Agent, Muse API and Muse Code.

Who is leading Meta's enterprise AI business?

Chirantan "CJ" Desai, formerly president and CEO of MongoDB, is Meta's chief enterprise platform officer and reports directly to Mark Zuckerberg.

Should enterprises buy AI from Meta?

It is worth evaluating, but Meta has not published enterprise pricing or detailed data commitments, so buyers should get written terms on training, ad separation and retention before piloting.

Filed under: AI Strategy & ROI, Agentic AI, AI Governance & Trust