When Meta went looking for someone to run its new enterprise AI business, it did not pick a research scientist. It picked Chirantan “CJ” Desai, the chief executive of the database company MongoDB, a hire that says as much about Meta’s intentions as the product itself.
Meta said on September 28 that it was launching the Meta Enterprise Platform, a standalone business that packages the company’s AI capabilities for companies and developers, and that Desai would run it. Reuters and Bloomberg reported that the platform will open a stack that includes Muse, Meta’s consumer assistant, along with Meta Business Agent, Muse API and Muse Code.
The bundle is the clearest sign yet that Meta wants to turn the momentum of a consumer product into a business line. Muse has become the company’s answer to ChatGPT, a general assistant that millions of people use to write, search and code. The enterprise platform takes the same technology and wraps it for the workplace, with tools for building agents and embedding them in a company’s own systems.
Muse arrived as a consumer assistant and quickly became the product Meta pushed into search, messaging and shopping. Opening the assistant to businesses through an API and a business-focused version turns a free consumer product into something with a price tag, and the company has signaled it will charge for the enterprise tools rather than run them as a way to feed the advertising machine.
The choice of Desai is the part competitors will notice. MongoDB sells databases to businesses, and Desai built his career in infrastructure and enterprise sales rather than in the labs where models are trained. A person close to the company said the selection was deliberate: Meta already has the researchers. What it needed was someone who knows how to sell to chief information officers.
Desai knows the buyers Meta now wants to reach. MongoDB’s business runs on reliability, support and long-term contracts rather than on engagement metrics, and its customers are the developers and technology officers who decide what software a company runs. That is the discipline Meta is buying, and the reason the hire came from a database company rather than a rival AI lab.
The move marks Meta’s first attempt to enter the enterprise software market as a platform, rather than as a collection of tools. The company has long made money from advertising, and its AI work has been aimed mostly at consumers and at its own internal systems. Standing up a separate business line with a hired chief executive from outside the company is a different kind of bet.
The timing is not accidental. The same week, Shopify opened its checkout to AI agents, including Meta’s Muse, a step that turned the assistant from a thing people chat with into a thing that can complete a purchase. Meta has been moving Muse toward action, and an enterprise platform is where that capability gets sold to businesses that want agents to work inside their own software.
Meta is not the first to the market, and it will not be the last. Microsoft has built an enterprise AI business around OpenAI’s models, Google sells its own through cloud contracts, and Salesforce and others have raced to add agents to their products. Meta’s advantage, analysts said, is a consumer brand with reach that rivals any of them, if it can convince businesses the same assistant people use at home belongs in the office.
The risk runs the other way too. Enterprise buyers demand reliability, support and the kind of service agreements that consumer companies rarely bother with, and Meta has no track record to point to. Desai’s job, in part, is to borrow the credibility of his old employer and attach it to a product that has to work inside banks, hospitals and factories.
The move fits a broader pattern. Consumer technology companies have repeatedly tried to walk upmarket into business software, and the results are mixed. What separates Meta’s attempt, analysts said, is that the underlying models are already paid for by the advertising business, so the enterprise line does not have to carry the cost of the research that produced them.
What Meta is really testing is whether a company built on advertising can sell software the old-fashioned way. The enterprise platform bundles technology Meta was already giving away inside its own products and asks businesses to pay for it. Whether that works will depend less on the quality of the models than on the sales channel Desai is being paid to build.
For a company that has spent two decades telling advertisers it understands people, the enterprise push is a bet that it can understand institutions too. The database chief from MongoDB is the first hire in that effort, and the platform he has been handed is the first product. Both will be judged on whether they bring in revenue, not just attention.
