Meta has announced a new enterprise AI business built around four parts of its existing AI stack: the Muse agent, Meta Business Agent, Muse API, and Muse Code. The company says it will turn these tools into products businesses and developers can deploy for their own operations. It has also appointed former MongoDB CEO Chirantan “CJ” Desai to lead the group as Chief Enterprise Platform Officer, reporting to Mark Zuckerberg.
The announcement sets a direction, but it does not describe a finished enterprise bundle. Meta has not published a platform price, a sales date, customer contracts, deployment options, service guarantees, or a full list of enterprise administration features. Several named products already have separate launches and availability. Meta Enterprise Platform is the new business umbrella intended to bring them to business customers. (Meta’s September 28 announcement)
Updated September 28, 2026. This explainer is based on Meta’s announcement and its existing product documentation.
What is Meta Enterprise Platform?
Meta Enterprise Platform is Meta’s new effort to sell AI products and services directly to organizations. Zuckerberg called it “the next major pillar” of Meta’s business. The company says the offering will draw on its models, agents, infrastructure, and existing work with businesses.
That matters because Meta’s business tools have largely helped companies reach customers through Facebook, Instagram, Messenger, and WhatsApp. The new platform points toward a broader role: Meta wants companies to use its AI to handle customer conversations, build their own AI features, and automate internal work.
The announcement names four starting components. They address different jobs rather than one single product:
| Component | Main job | What Meta has said so far |
|---|---|---|
| Muse agent | Carry out multi-step tasks for a person | A personal agent that can plan, use connected apps and a browser, and continue work in the background; Meta has not described its enterprise configuration |
| Meta Business Agent | Respond to customers and help complete business tasks | Answers business questions, recommends products, books appointments, qualifies leads, hands off to staff, and can close sales |
| Muse API / Meta Model API | Let developers use Meta models in their own software | Meta says the Model API is generally available globally, with enterprise workloads supported through its API and cloud partners |
| Muse Code | Help engineers work through software changes | A coding agent that uses parallel workers in isolated workspaces for longer coding tasks |
Meta has not said these pieces will be sold together, share one management console, or use one set of terms. “Platform” describes the business direction; the exact product architecture is still to come.
1. Muse: an agent that can take actions
Muse is Meta’s personal AI agent. Instead of only answering a prompt, it can break a goal into steps and work across connected services. Meta’s examples include sending email, booking travel, filling out forms, and continuing a longer task after the user leaves the app. It can ask for approval before sensitive actions such as sending a message or making a purchase. (Meta’s Muse overview)
Meta says Muse runs inside a dedicated cloud virtual machine with its own browser and the user’s connected data. A separate “Sentinel” agent checks its internet actions. Users can choose which apps to connect and what permissions to grant, review an audit trail, disconnect services, and opt out of having interactions used to train Meta AI models. Meta says Muse conversations and VM data are not shared with its advertising systems. Those are Meta’s statements about Muse as announced; business customers will need the enterprise product terms to establish what protections and controls apply to their deployments.
For organizations, an agent like Muse could take on repeatable work that spans several tools: gather information, draft a response, update a record, and ask a person to approve the final action. Meta has not yet explained how companies would configure team-level agents, connect private business systems, manage identities, or monitor usage through Meta Enterprise Platform.
2. Meta Business Agent: customer service in Meta’s messaging apps
Meta Business Agent focuses on the point where companies already meet customers: chat. Meta says businesses can configure agents to answer questions about their products and policies, recommend items from a catalog, book appointments, qualify incoming leads, route a conversation to a human, and complete sales.
The related Meta Business Agent Platform gives larger businesses infrastructure to build, customize, and deploy agents at scale. Meta says it can connect to hundreds of systems, including Shopify, Zendesk, and Shopee, so an agent can take actions using business data and tools. The company also says the platform includes controls, guardrails, and measurement for businesses to set rules and personalize customer experiences. WhatsApp is a primary surface, with Messenger and Instagram supported too. (Meta Business Agent announcement)
This is the most direct bridge between Meta’s current business and its enterprise AI ambitions. Meta already operates the messaging channels where many customer conversations happen. Business Agent adds an AI layer that can answer and act inside those conversations. Meta has not yet detailed enterprise pricing, exact system coverage, regional availability by feature, or the process for evaluating the agent’s answers and actions before a company puts it in front of customers.
3. Muse API: models developers can build with
Zuckerberg’s launch post calls this the “Muse API.” Meta’s developer material refers to the developer offering as the Meta Model API. It gives developers access to Meta models so they can build AI into their own apps and services rather than sending employees or customers to a Meta chatbot.
At Meta Connect on September 24, Meta said the Model API was generally available worldwide. The same update described Muse Spark 1.3 as suited to complex coding and agentic workflows, Muse Image for image generation, and voice tools for transcription. Meta said enterprise workloads are supported through its first-party API and cloud partners. It also named availability for Muse Spark 1.3 on Oracle Cloud AI Platform and a private preview on Google Cloud. (Meta for Developers’ Connect recap)
For a company, an API can be the most flexible component: teams can choose the user interface and workflow, while calling Meta’s models for reasoning, text, image, or audio tasks. The trade-off is that the customer usually has to build or integrate more of the surrounding application, including access controls, logging, evaluation, and a reliable process for human review.
4. Muse Code: a coding agent for engineering teams
Muse Code is Meta’s coding agent. Meta describes it as a tool for substantial, multi-step engineering work. Its developer overview says it is multi-agent by default: parallel workers can write and test code in separate workspaces. Meta said Muse Code was out of beta and available on Windows at Connect.
That gives Meta a way to sell more than model access. Coding agents include a working environment and orchestration around a model: they inspect a codebase, make edits, run tests, and coordinate multiple steps. Meta has not said how enterprise teams will centrally manage Muse Code, what repository and identity integrations are supported, or how enterprise subscriptions and usage limits will work under the new platform. (Meta for Developers’ Connect recap)
How the pieces could fit together
The four components suggest a stack. Meta’s models provide the underlying generation and reasoning. Muse and Business Agent package those models into agents for task completion and customer conversations. Muse Code packages them for software development. The API lets a company or software vendor build a different experience around the models.
A retailer, for example, might use Business Agent to answer order and product questions in WhatsApp, and connect it to its catalog and support system. Its engineering team might use Muse Code to work on the storefront. If the retailer wanted an AI assistant inside its own app, developers could use the Model API to build one. These are examples of how the announced components might be used; Meta has not announced this as a single integrated customer workflow.
Security, privacy, and the open questions for buyers
Meta says security and privacy are built into its enterprise products from the outset. That is a commitment, not a substitute for product documentation or contract terms. The Muse product announcement describes controls such as permission choices, approval prompts, audit history, and a dedicated virtual machine. The Business Agent announcement describes enterprise controls and guardrails. The September 28 announcement does not spell out how those features map to a company’s compliance program or procurement requirements.
Before adopting the platform, buyers will want answers about data retention and training, data residency, encryption and key control, single sign-on and role-based access, audit exports, model and agent evaluation, human escalation, incident response, uptime commitments, support, indemnity, and pricing. Meta has not provided those details in the launch announcement.
Availability is also split across products. Meta says the Model API is generally available globally and Muse Code is out of beta. Meta Business Agent is expanding across its messaging products, with paid subscriptions previously described as forthcoming. Muse itself has its own consumer rollout. None of those statements establishes that a complete Meta Enterprise Platform package is available to buy today.
Who should pay attention?
Customer support leaders should watch Meta Business Agent because it puts AI directly into channels their customers may already use. Developers and product teams should look at the Model API if they want to test Meta models in their own software. Engineering leaders can evaluate Muse Code as a coding agent. Operations teams may be interested in agents that can take actions across connected services, but they will need clear permission and audit controls before assigning real work.
Desai’s appointment gives the effort an executive owner with experience in enterprise software. Meta says he was CEO and president of MongoDB, led product and engineering at Cloudflare, and spent nearly eight years at ServiceNow, including as president and COO. He will report directly to Zuckerberg.
The next useful evidence will be specific: a product catalog, enterprise pricing, availability dates, customer references, admin documentation, and the contractual controls behind Meta’s security claims. Until Meta publishes those, buyers can assess the individual products, but they cannot yet compare a defined Meta Enterprise Platform package with competing enterprise AI contracts.
Frequently asked questions
Is Meta Enterprise Platform a new AI model?
No new model was announced as part of the platform launch. Meta is bringing existing models and products, including the Muse agent, Business Agent, the model API, and Muse Code, to businesses and developers.
Can companies buy Meta Enterprise Platform today?
Meta has not announced a general availability date, one platform price, or a complete enterprise package. Some underlying products are already available through their own channels, but availability varies by product.
How much will it cost?
Meta has not published Meta Enterprise Platform pricing. Pricing for individual APIs, tools, or Business Agent plans may be announced separately and should not be treated as a platform price.
Who will lead Meta Enterprise Platform?
Chirantan “CJ” Desai will join Meta as Chief Enterprise Platform Officer and report to Mark Zuckerberg. Meta says Desai is leaving MongoDB, where he was CEO and president.
Sources
– Meta: Launching Meta Enterprise Platform – Meta: Be There for Every Customer With Meta Business Agent – Meta: Introducing Muse – Meta for Developers: Meta Connect 2026 recap – Mark Zuckerberg’s announcement on X
The Kingy Brief
Get the next Kingy Brief.
Source-checked AI changes, original tests and one practical thing to try.
Free · Choose your subjects · Double opt-in · Unsubscribe anytime
After submitting, check your inbox for "Confirm your subscription to The Kingy Brief" and open its confirmation link. Check Spam or Promotions if you cannot find it. Regular editions remain paused; no restart date is set.