A few months ago, Meta looked like an AI underdog. Now its Muse app has topped the U.S. iPhone charts, its underlying model is improving quickly, and Mark Zuckerberg is calling enterprise AI “the next major pillar” of the company’s business.
That is quite a change in momentum. Muse has become a hit with people well beyond the usual AI crowd by offering something easy to grasp: message an agent about a task, and it can keep working after you close the app. The personal agent reached No. 1 on the U.S. free iPhone chart, ahead of ChatGPT. Now Meta wants to bring the technology behind that consumer breakout to companies.
In announcing Meta Enterprise Platform, Zuckerberg wrote, “We believe superintelligence will create significant new opportunities for all people and businesses.” The initial pitch brings together Muse, Meta Business Agent, Muse API, Muse Code, and the company’s broader AI stack for businesses and developers.
From a viral agent to an enterprise platform
The names are easy to tangle. Muse Spark is the model. Muse Code is the coding agent built on it. Muse is the personal agent people download. Meta Enterprise Platform is the new business effort meant to turn that growing stack into products companies can use.
Meta has updated Muse Spark repeatedly since its April debut. The September release, Muse Spark 1.3, focuses on longer tasks, coding, tool use, and following detailed instructions. Meta says that, in comparisons by its own engineers, the new model used roughly 20% fewer tool calls and 25% fewer tokens than its predecessor. Those are company figures, but they point to a practical goal: an agent that can finish a job without wasting time and money along the way.
Muse Code puts the model to work in software development. Launched in beta in August, it can plan changes across large repositories, write code, and validate results. Meta says its persistent background agents and restart-safe event log help it handle work that runs longer than a quick coding prompt. At Connect, Meta said Muse Code was out of beta and available on Windows.
The new platform extends that developer push into a broader business proposition. Zuckerberg said Meta intends to draw on strengths “few other companies have,” including models, agents, infrastructure, and years of working with businesses. Meta already helps hundreds of millions of businesses reach customers across its products. Its bet is that those relationships can become a starting point for selling AI that helps them operate and grow.
A new leader, and a different kind of customer
Zuckerberg has hired Chirantan “CJ” Desai to lead the effort as Chief Enterprise Platform Officer, reporting directly to him. Desai was most recently CEO and president of MongoDB; earlier, he led product and engineering at Cloudflare and held senior roles at ServiceNow.
Desai described the job plainly: Meta Enterprise Platform will focus on “turning its AI stack into products and services that companies can deploy for their own businesses.” He also said security and privacy are being built into those products from the outset, and set an ambitious goal: “make Meta the place enterprises come to scale their businesses.”
That goal will take more than a chart-topping app. Business customers will judge Meta on reliability, privacy, security, cost, and how well its tools fit existing workflows. Meta has made its Model API generally available globally and says it supports enterprise workloads through its own API and partners. Muse Spark 1.3 is available through Oracle Cloud AI Platform, with Google Cloud availability in private preview. Those routes give developers ways to test Meta’s claims in the environments where they already work.
Still, the timing is hard to ignore. When Muse Spark first arrived, The Neuron described Meta as back in the AI race. Five months later, the company has improved the model, shipped a coding agent, and put a personal agent into the hands of people who may never have followed an AI benchmark.
Muse has given Meta attention. Meta Enterprise Platform is its attempt to turn that attention, along with its models and enormous business reach, into a durable new business. The next question is whether companies find its AI as compelling to rely on as consumers have found Muse to try.
Related reading
- How to get started with Meta Muse explains what the personal agent does and where it fits.
- Gemini 3.8 and Muse Spark 1.3 go head to head examines the model’s progress and the cost of completing real tasks.
- Meta’s big AI bet now fits on your keychain covers the wider Muse ecosystem announced at Connect.
