What To Know
- He told the Connect audience that Muse was the centerpiece of the company’s vision and predicted that it could eventually develop into a form of personal superintelligence used by billions of people.
- Meta also intends to put Muse directly into its smart glasses, potentially making the agent available without users needing to reach for a smartphone or sit in front of a computer.
Meta’s newly launched Muse is rapidly emerging as one of the most closely watched consumer AI agents, with the company racing to expand its capabilities only weeks after its debut. From controlling computers and working through email to shopping, booking appointments, making calls, and eventually operating through smart glasses, Meta is positioning Muse as something considerably more ambitious than another conversational AI assistant.
Image Credit: Thailand AI News
The scale of that ambition became clearer at Meta’s Connect event in Menlo Park, where CEO Mark Zuckerberg placed Muse at the heart of the company’s artificial intelligence strategy. Meta wants the agent to move beyond answering questions and actually carry out jobs across the apps and services people already use. In the middle of that rapidly developing picture, this AI Agents news report looks at how Muse is evolving into a digital worker that could remain active across a user’s devices, accounts, and everyday activities.
Muse arrived in early September as a personal AI agent powered by Meta’s Muse Spark multimodal model, which was specifically developed for agentic work. Unlike a traditional chatbot that generally waits for a question and generates an answer, Muse is designed to work through multi-stage tasks and interact with external services when users give it permission.
That could mean arranging travel, sending emails, buying products, handling administrative work, searching for information, or performing other tasks that would normally require users to jump between several websites and applications themselves.
Zuckerberg has made clear that Meta’s plans extend considerably further. He told the Connect audience that Muse was the centerpiece of the company’s vision and predicted that it could eventually develop into a form of personal superintelligence used by billions of people.
Meta is also thinking about how that activity could eventually become a business. Zuckerberg said Muse would initially be free for a large number of tokens, while suggesting Meta could ultimately earn money by taking a small fee from transactions conducted through the agent.
Meta Gives Muse a Digital Face
One of the more eye-catching developments announced at Connect involves transforming Muse from an invisible software assistant into an AI character users can see and speak with.
Meta has developed Muse Realtime Avatar, which will allow users to create a digital avatar for their individual agent and conduct real-time video conversations with it.
The concept gives Muse a face, body, and voice rather than restricting interactions to a conventional chat interface. Users could speak directly to their avatar, ask questions, issue instructions, and assign jobs while interacting with something that feels more like a digital character.
Zuckerberg identified the physical avatar demonstrated during the event as Jolly.
There are obvious echoes of Meta’s longstanding interest in virtual worlds, but the difference is that the avatar is being attached to an AI agent designed to perform practical tasks. Instead of simply representing a user or character inside a virtual environment, it becomes the visible personality of software capable of taking action.
Muse Is Coming to Meta’s Smart Glasses
Meta also intends to put Muse directly into its smart glasses, potentially making the agent available without users needing to reach for a smartphone or sit in front of a computer.
The planned integration will allow people to activate Muse through a wake word and speak instructions through their glasses.
Meta says the agent could guide someone through a personalized workout, record meals, arrange an appointment, or help purchase products that the wearer encounters.
Perhaps more significantly, Muse is designed to continue working after receiving an instruction. Rather than forcing someone to remain inside an app while a task is completed, the agent can operate in the background and return when the job is finished.
Meta has not provided a firm release date for the smart-glasses functionality, saying it is expected in the coming months.
Muse Wants to Take Control of the Computer
Another major step involves bringing Muse’s computer-working capabilities to Mac.
This would allow users to delegate tasks that require the AI to interact directly with desktop applications. The significance is considerable because it moves Muse beyond providing instructions about how to accomplish something and toward actually operating the software needed to finish the job.
Meta Chief AI Officer Alexandr Wang said the technology could help people run a small business or simply complete ordinary work.
His description of the feature also illustrates the broader promise of agentic AI: a user could line up several jobs, leave the computer, and allow Muse to continue working independently.
Computer control is becoming an important battleground among AI developers because it potentially removes one of the biggest barriers facing autonomous agents. Instead of requiring specially designed integrations for every service, an agent capable of operating existing applications can theoretically work with software in much the same way a person does.
Muse Will Have Its Own Email Address
Meta is also preparing to give Muse an email address of its own.
The planned feature could make the agent a more active participant in users’ communications. People will be able to include Muse in an email conversation or forward messages to the agent and ask it to handle related tasks.
That could turn email itself into another command channel for AI.
An itinerary, invoice, meeting request, customer query, or other message could potentially be forwarded to Muse with instructions about what should happen next.
Meta says the feature will arrive soon, although the company has not provided a precise timetable.
Shopping Could Become One of Muse’s Biggest Jobs
Meta is making an aggressive push to connect Muse with commerce, retailers, and payment providers.
Announced relationships and integrations involve Shopify, Stripe, PayPal, Expedia and Instacart, alongside retailers including Best Buy, Gap, Sephora, Walmart, and Wayfair. Meta has also announced connectors involving services such as GitHub, Granola, and Notion.
Through Shopify, Muse can access a huge catalog of products, while payment integrations are intended to make it possible for the agent to progress from finding an item to helping complete the purchase.
This could dramatically change online shopping if consumers become comfortable allowing AI agents to search across products, compare options, and carry out transactions.
There are limits, however. Amazon has reportedly blocked Muse from shopping through its platform. The company argued that unauthorized AI-agent access violated its terms and raised concerns involving user credentials and account information.
Meta, meanwhile, wants many more companies to connect directly with Muse. Wang said more than 1,500 applications were received in less than a week after developers were invited to build connectors for the platform.
Early Users Are Finding Surprising Uses for Muse
Perhaps the strongest indication of Muse’s potential comes from the variety of jobs early users say they are already giving it.
Reported examples range from obtaining airline compensation and dealing with customer-service calls to canceling subscriptions, planning a house move, finding restaurants, monitoring travel prices, researching insurance, removing information from data-broker websites, and putting together outfits.
One traveler reported receiving a $250 airline credit after asking Muse to deal with the consequences of a seven-hour flight delay. The agent reportedly investigated alternative flights and handled parts of the support process.
Another user described having Muse call an airline and remain on hold until a human representative became available, allowing the customer to avoid spending the time waiting.
Other reported users have employed Muse to identify unwanted subscriptions, with one saying the exercise saved $200. Another user reportedly saved $600 after using Muse to monitor already-booked flights and hotels for price reductions.
These are individual accounts rather than evidence that every Muse user will achieve similar results. Nevertheless, they demonstrate why personal agents are attracting attention: the most appealing applications may involve tedious tasks people understand perfectly well but simply do not want to spend their time completing.
Rapid Adoption Gives Meta an Early Boost
Muse has also generated substantial early download activity. According to Apptopia figures cited in the supplied reports, Muse recorded around 1.8 million iOS downloads in the United States and Canada during its first 12 days. ChatGPT recorded approximately 1.3 million downloads during the comparable period following its mobile release.
Including Android, total Muse downloads reportedly reached approximately 2.8 million.
Engagement figures were also strong. Apptopia estimated Muse had 642,000 daily active users during its first 12 days, compared with 231,000 for ChatGPT during its corresponding post-release period.
Separate Sensor Tower figures cited by Reuters placed Muse downloads above 2.5 million shortly after launch. Differences between estimates can reflect the methodologies and measurement periods used by analytics companies.
Meta also possesses something few AI competitors can easily replicate: an enormous existing consumer network.
Apptopia reported that 95 percent of early Muse adopters were Facebook users, while 63 percent also used Instagram. Meta can therefore introduce its AI agent to people already participating in its broader ecosystem rather than having to build an audience entirely from scratch.
The Human Concierge Experiment Raises Privacy Concerns
Muse’s rapid development has not been without controversy. It was reported that Meta had been experimenting with a “human concierge” arrangement under which contractors could quietly handle some telephone calls initiated through Muse.
The system was reportedly tested as Meta attempted to improve the success rate of telephone tasks. Some businesses and customer-service representatives have been reluctant to interact with automated callers, creating a practical obstacle for AI agents expected to complete real-world jobs.
However, Meta employees reportedly raised concerns that people could inadvertently disclose sensitive information to contractors if they believed an AI system was handling the entire conversation.
Accordingly, Meta rolled back the feature for the time being. A company executive acknowledged internally that testing contractor-assisted calls without adequate disclosure had been a mistake.
Meta said it was using testing to develop appropriate safety and privacy protections and indicated that any public rollout would require proper disclosure.
The episode illustrates a broader challenge facing autonomous AI. As agents become capable of acting on behalf of users, transparency about how those actions are performed becomes just as important as whether they can be completed successfully.
The Real AI Agent Race Is Only Beginning. Muse’s explosive arrival shows how quickly consumer artificial intelligence is moving beyond the chatbot era. The next competitive frontier is increasingly about AI systems that do things rather than simply tell people how to do them.
Meta wants Muse to operate across computers, email, commerce, travel, payments, smart glasses, communications, and potentially countless third-party services. If that strategy works, the agent could become an intermediary between users and a significant portion of their digital lives.
That also raises the stakes. An agent with access to email accounts, payment services, shopping platforms, computers, calendars, and personal information must be reliable, secure, transparent, and predictable. Mistakes become considerably more consequential when AI is authorized to take action rather than merely generate a response.
Muse’s early adoption, expanding list of integrations, and rapidly developing capabilities suggest that Meta intends to compete aggressively for a central position in the emerging AI-agent economy. Whether consumers ultimately trust an autonomous assistant with increasingly important parts of their lives remains an open question, but Muse’s first weeks demonstrate that there is considerable interest in AI that can move beyond conversation and actually get work done.
For more on Meta’s Muse, visit:
