The short answer
Meta is now selling its AI to businesses: a new unit, Meta Enterprise Platform, bundles the Muse agent, Meta Business Agent, the Muse API and Muse Code, and is run by former MongoDB CEO CJ Desai. For companies planning generative AI integration, it adds a fourth large vendor to the shortlist next to OpenAI, Anthropic and Google, but without published pricing or enterprise terms yet.
The move matters less for what ships today than for what it signals: Meta is treating enterprise AI as, in Mark Zuckerberg’s reported words, “the next major pillar” of its business. Expect aggressive pricing and bundling with WhatsApp, Messenger and Instagram once commercial terms arrive.
What did Meta announce?
Meta created a dedicated business unit to sell its AI stack to companies and developers. According to Meta’s announcement, Meta Enterprise Platform will offer “advanced AI models and infrastructure” alongside four products it already runs: the Muse agent, which can carry out tasks such as sending email or booking travel; Meta Business Agent, which answers customer questions, recommends products, books appointments and qualifies leads across Meta’s messaging apps; the Muse API for developers; and Muse Code, Meta’s coding tool.
The unit is led by CJ Desai, who left MongoDB the same day. Desai said Meta brings together “advanced models and leading agents with a proven track record of helping millions of advertisers and hundreds of millions of businesses scale.” TechCrunch reported that MongoDB shares dropped more than 17% on the news; Bloomberg and CNBC also covered the move.
What Meta did not say is as important. Neither the announcement nor the press coverage lists prices, availability dates, service levels or enterprise admin controls. VentureBeat described the launch as “a strategy more than a purchasable package” and noted that a confidential virtual machine designed to keep Meta out of agent workloads is still forthcoming.
Why is Meta moving into enterprise AI now?
Enterprise contracts are where AI revenue is concentrating, and Meta has so far been absent while OpenAI, Anthropic and Google compete for corporate budgets. Hiring a chief executive who ran enterprise sales and product at MongoDB, Cloudflare and ServiceNow signals that Meta intends to build a proper B2B organization rather than expose a model endpoint and wait.
Meta also has an asset the model labs lack: direct reach to hundreds of millions of businesses already on WhatsApp, Messenger and Instagram. Meta Business Agent is the obvious wedge. For retailers and service companies whose customers already message them on those apps, a first-party agent inside the channel will be hard to ignore.
What it means for US & EU software teams
More competition should push prices down, but only for teams that can switch. A new vendor with a large distribution business is likely to compete on price. Teams whose code calls one provider’s SDK directly, with prompts and tool definitions tuned to that model, will struggle to take advantage. A thin internal abstraction for model calls, tool schemas and evaluation lets you benchmark Meta against incumbents with real traffic instead of marketing claims.
Channel agents raise integration questions, not only model questions. A Business Agent that books appointments or qualifies leads needs reliable access to your inventory, calendar, CRM and order data. The hard work is in those APIs, permissions and fallbacks to a human, and it is the same whichever vendor supplies the model.
EU teams need terms before pilots. Meta’s consumer AI products have faced scrutiny from European regulators over training on user data. Before any customer data flows into a Meta agent, EU companies should have a data processing agreement, clarity on training use and data location, and audit logs they can export. Without them, a pilot can create GDPR exposure that outlasts the experiment.
What should enterprise buyers ask Meta?
- Pricing and commitments. Per-token, per-seat or per-conversation pricing, minimum spend and how prices change after any launch discount.
- Data use. Whether prompts, files and agent activity are used to train Meta models, and how to opt out contractually.
- Data location and isolation. Where workloads run, whether EU processing is available, and when the confidential VM ships.
- Admin and audit. Single sign-on, role-based access, approval workflows for agent actions and exportable logs for security teams.
- Exit path. Model and API deprecation policy, and whether agent configurations can be exported if you leave.
Frequently asked questions
What is Meta Enterprise Platform?
Meta Enterprise Platform is a new Meta business unit, announced on September 28, 2026, that packages Meta's AI models and agents for companies and developers. At launch it groups four offerings: the Muse agent, Meta Business Agent, the Muse API and the Muse Code coding tool. Meta says security and privacy are built into its enterprise products from the outset.
Who leads Meta's enterprise AI business?
Chirantan "CJ" Desai, who stepped down as MongoDB CEO on September 28, 2026 after about 11 months in the role. He is Meta's chief enterprise platform officer and reports directly to Mark Zuckerberg. Before MongoDB he led product and engineering at Cloudflare and was president and COO of ServiceNow. MongoDB named Dev Ittycheria interim CEO.
How much does Meta Enterprise Platform cost?
Meta has not published pricing or availability dates. Meta's announcement and TechCrunch's report both describe the unit and its products but give no commercial terms, so teams cannot yet compare total cost against OpenAI, Anthropic or Google enterprise offerings.
Should my company switch to Meta's AI agents?
Not yet as a primary platform. The announcement describes a strategy more than a purchasable package: pricing, admin controls, audit logging and contractual data terms are not public. It is reasonable to add Meta to your vendor shortlist and to design your application so the model and agent provider can be swapped, then evaluate once terms are published.
What should EU companies check before using Meta's enterprise AI?
Ask where prompts, files and agent activity are processed and stored, whether customer data is used for model training, what data processing agreement is offered, and whether audit logs and deletion controls exist. These answers determine whether a deployment can meet GDPR obligations and, for relevant use cases, EU AI Act logging and human-oversight requirements.
Sources
Meta Newsroom — Launching Meta Enterprise Platform
TechCrunch — Meta launches enterprise AI platform, hires MongoDB CEO to lead new initiative
CNBC — Meta hires MongoDB CEO CJ Desai to lead enterprise unit
Bloomberg — Meta taps MongoDB CEO to lead new enterprise AI platform
VentureBeat — Meta announces enterprise AI platform, recruits MongoDB CEO to lead it
SiliconANGLE — Meta hires MongoDB CEO CJ Desai to lead new enterprise AI business