By Amara Osei

Meta has hired former MongoDB CEO Chirantan "CJ" Desai to build a new enterprise platform, extending its Muse personal agent beyond consumers and advertising tools into one of the most contested markets in business software. The company also announced Muse for Small Business, which will integrate the agent into tools including Box, Figma and Shopify.

Details on the full enterprise platform remain vague. Muse Code and Meta Business Agent will count among the existing offerings. That gap between announcement and architecture matters: Meta has not said what the platform will actually do, who it is priced for, or when it will ship.

What does Meta bring to the enterprise market?

Forrester principal analyst Kate Winick said Meta already functions as a growth partner for the millions of businesses using Meta Ads Manager, where the company has rolled out AI to accelerate creative production and audience targeting — to mixed reactions from advertisers.

"The question of whether or not you personally trust Meta with your data is a different question than whether or not your company runs business through Meta," Winick said. "For most businesses, the answer is already yes. And so this allows [Meta] to sort of bring together the frontier model that they've built with the enterprise relationships they already have."

The claim deserves stress-testing. Running ads through Meta is not the same as connecting an AI agent to QuickBooks, Stripe, customer lists and email. Jeremy Schulkin, SVP of services at marketing agency Hawke Media, framed the risk in direct terms: "We have fought hard to not hand over the keys to Meta. Connecting QuickBooks, Stripe, your customer list, and your email gives Meta a full view of your business."

Schulkin also flagged a structural conflict: Meta effectively grades its own work. An agent built by the company selling the ads has little incentive to advise a business that its budget would perform better on TikTok.

How well does Muse actually work?

Early users report genuine capability — with limits.

• Jack Johnston, VP of social innovation and growth at Tinuiti, said the agency has piloted Meta's model context protocol (MCP) server (a standard for connecting AI agents to external tools). The agentic campaign manager in Muse for Small Business leans heavily on it and suits small businesses unfamiliar with Meta ads. "Comparing it to enterprise or SMB advertisers that already operate in Ads Manager, it can be a useful guide, but it lacks some of the sophistication that may be required," he said, adding that he expects Meta to add sophistication as it moves beyond small businesses.
• Kevin Gosschalk, founder and CEO of cybersecurity firm Arkose Labs, called Muse far and away the best enterprise agent he has used in terms of computer use (the ability of an AI agent to operate software the way a person does). Arkose's web data shows more traffic driven by Muse since it gained traction with consumers.

But Gosschalk's endorsement comes with a hard condition. Arkose's security team is reviewing Muse for enterprise use, checking whether it would comply with the strict data requirements of clients such as financial institutions. "Muse currently is a cloud-based service. So anything you give it goes into Meta's data centers. That, we're not able to do generally in a workforce environment," he said. He wants Meta to ship a locally hosted version, like its latest models.

Meta spokesperson Kamran Mumtaz pointed to a company blog post stating that Meta designed Muse with "multiple layers of privacy and security protections." The statement does not address local hosting or enterprise compliance directly.

Why the skepticism runs deeper than the product

Meta has tried workplace software before. Its communication platform Workplace, launched in 2016, was shuttered in May of this year.

Winick drew a sharp line between that failure and the current push: "It was groups, it was internal messaging, and frankly, internal comms is a much lower priority for most businesses than growth and revenue. This is targeted at a much more critical part of the business and working on much bigger problems than Workplace was ever intended to address."

Diego Lomanto, CMO of AI enterprise startup Writer — a competitor to Meta's new platform — said Meta must demonstrate "separation between church and state" between its consumer-data-driven ads business and the enterprise platform before companies will trust it with their data.

The competitive clock is already running. Rivals have spent years entrenching their own enterprise AI, and Meta is entering late without a cloud arm or office software suite. "They're going to have to make a strong case for why people need to switch or why the ad partner is the right person to build the overall business infrastructure," Winick said. "They have relationships and they have a way in, but it's not going to guarantee them the sale."

Meta's relationships open the door; whether security and neutrality concerns let it close the deal is the question every prospective enterprise buyer will ask.