The artificial intelligence race is entering a more personal phase.
For the past several years, the technology industry has focused on chatbots that answer questions, summarize documents and generate images. Meta now wants consumers to hand AI something more valuable: permission to act.
On Sept. 8, Meta launched Muse, a personal AI agent designed to work across the apps and services people use every day. The company says Muse can send emails, book travel, fill out forms, negotiate on a user’s behalf and help turn long-term goals into action plans.
That makes Muse more than another chatbot. It is designed to continue working after the user closes the app, returning when something changes or when it needs approval before taking a sensitive action.
Meta is rolling out Muse in the United States through a dedicated app, its website and WhatsApp. The basic version is free, while heavier users can pay $20 or $100 per month for additional usage.
The launch gives Meta a direct consumer product for its AI ambitions. It also creates the first clear test of whether the company can turn its massive infrastructure spending into a new source of revenue.
Meta needs more than better recommendations
Meta’s advertising business remains the financial engine behind the company. Facebook and Instagram generate enormous revenue by predicting what users will watch, click and buy.
Muse represents a different opportunity. Instead of simply recommending a product, it could eventually help select the product, complete the purchase and manage the entire process.
Meta says Muse can use Stripe’s Link payment system to check out with a one-time-use card, keeping the user’s actual card details hidden. Additional integrations, including Shop Pay and 1Password support, are expected later.
That model could create several potential revenue streams. Meta could collect subscription revenue, participate in commerce transactions or use the agent to make its broader collection of apps more valuable.
However, the subscription math shows why Muse is unlikely to justify Meta’s AI spending on its own in the near term.
If 10 million users paid $20 per month, Muse would generate $2.4 billion in annualized gross subscription revenue before costs, churn and payment fees. That would be a meaningful business, but it would still be small compared with Meta’s expected 2026 capital expenditures of $130 billion to $145 billion.
The larger financial argument is that Muse may increase the value of Meta’s entire ecosystem. A successful agent could make WhatsApp, Instagram and Facebook more useful while creating new opportunities in shopping, services and digital payments.
The infrastructure bill is already enormous
Meta has been spending aggressively to build the computing capacity required for advanced AI.
The company’s latest guidance calls for $130 billion to $145 billion in 2026 capital expenditures, including principal payments on finance leases. Much of the spending is connected to data centers, servers and other infrastructure needed to train and operate AI systems.
That investment is happening before Meta has demonstrated how much revenue its consumer AI products can produce.
The company is effectively building the factory before proving the final product. That strategy can work if demand arrives quickly, but it also creates pressure on management to show that the spending is generating more than improved engagement or better advertising performance.
Muse is therefore important even if its first-year subscription revenue is modest. It gives investors a visible product that can be evaluated through paid conversion, user retention, commerce activity and the amount of work users actually delegate to the agent.
Trust may be the real product challenge
Meta is presenting Muse as a secure and private system. Each agent operates inside its own virtual machine, and users decide which apps and services it can access.
Meta also says a separate monitoring system can review planned actions and require authorization in certain situations. The company plans to introduce a more heavily encrypted version later this year.
Those safeguards are necessary because an AI that can read email, access calendars and make purchases creates much greater risks than an AI that only produces text.
Reuters reported that internal testing produced mixed results. Some employees found Muse useful for arranging complex travel, while others reported login problems, unreliable monitoring and incidents involving unauthorized exposure of sensitive data.
Those reports do not prove that Muse is unusable. They do show why consumers may hesitate before connecting their most important accounts to a new system that can operate in the background.
A personal AI agent does not need to be perfect to be useful. But it needs to be predictable. One incorrect email, unauthorized purchase or privacy failure could damage confidence quickly.
Investors are watching the behavior, not the demo
Meta’s launch is strategically significant because the company has distribution few competitors can match. WhatsApp, Instagram and Facebook already occupy a large part of billions of people’s daily lives.
The question is whether that reach can overcome the trust problem.
Consumers may try Muse because it is convenient, but continued use will depend on whether it reliably saves time without creating new problems. Meta must also convince users that their personal data will not become another source of uncertainty.
For investors, the most important numbers will not be the number of people who download the app. They will be the number of people who return, pay, connect additional services and allow Muse to complete increasingly valuable tasks.
Meta has made a large, expensive bet that the next major AI interface will not be a search box or a chatbot. It will be a digital assistant that quietly works in the background.
Muse is the company’s first serious attempt to make that future tangible for ordinary consumers. The opportunity is enormous, but the financial test is equally clear: Meta must prove that people will trust an AI agent enough to pay for it, use it regularly and let it participate in their real lives.
