本文来源：https://www.163.com/dy/article/L6CC3F8H05198NMR.html
若有版权问题，请来信告知：13951431913#139.com 或在右方提交评论。
0
Meta Platforms has officially launched Muse, an AI agent designed for individual users. This is not only another substantive step in Mark Zuckerberg's vision of integrating AI into everyday life, but also marks the company's major move in the consumer-grade AI assistant space. Muse opened to users in the United States on Tuesday and can handle a variety of everyday tasks on behalf of users, including online shopping, movie ticket bookings, sports class reservations, and even filling out school activity permission forms. Meta's Chief AI Officer, Alexandr Wang, pointed out that the core design philosophy of the product is to "make it easy for ordinary users to navigate." In a lengthy post of more than 6,500 words published last month, Zuckerberg clearly stated that consumer-facing personal AI assistants represent Meta's biggest business opportunity, and outlined this vision: "Everyone will have a capable personal agent that understands you, your goals, and everything you care about." Meta's ultimate vision is to have Muse reach its billions of existing users across platforms such as Instagram, Facebook, and WhatsApp. Meta's stock closed down 0.53% on Tuesday. Core Features: From Task Execution to Goal Planning Muse was designed from the start to be an agent tool that requires no technical expertise. Users can issue instructions through the dedicated Muse app or directly in WhatsApp using conversational commands, and Muse will automatically complete the operations — including launching a browser, filling out forms, sending emails, and even negotiating on the user's behalf. The product is powered by Meta's latest flagship model, Muse Spark, which is specifically designed for autonomous tasks in real-world scenarios. For time-consuming tasks, Muse can continue running even after the user leaves the app, and will promptly notify the user when there is new progress or when confirmation is needed — for example, before sending an email or completing a payment. Muse also features cross-app memory and association capabilities. If a user saves a recipe video on Instagram, Muse can automatically generate a shopping list; if a user mentions a friend's dietary preferences, Muse will take that into account when sending out dinner invitations. Meta defines this feature as proactive awareness that "remembers key information after just one mention." Architecture and Privacy: Cloud-Based Operation, User Data Accessible Only to the User In terms of privacy and security architecture, Meta has developed a virtual machine platform for Muse called Muse Secure VM, which ensures strict isolation between user data. Meta also revealed that later this year it will launch Muse Confidential VM, at which point user data will be encrypted with keys held solely by the user, making it unreadable even to Meta. Wang called it "an extremely secure architecture" — Muse stores operation logs and pending execution plans, does not share any data with Meta's advertising platform, and proactively seeks user confirmation before performing sensitive operations. For non-sensitive matters, Muse can handle them on its own. Wang named his own agent Euler (after his pet dog) and uses it to select and purchase furniture for his new home, arrange meal plans, and develop fitness programs. "I really do use it as a second brain," he said. Users can customize their agent's name and avatar, set its communication style, and connect apps such as email and fitness trackers so it can continuously advance personal goals in the background. Ecosystem Integration: From Ticketmaster to the Google Suite Muse has already connected with multiple major service platforms, including Ticketmaster, OpenTable, and Google's Gmail, Google Calendar, and Google Drive. At the same time, Muse is deeply integrated with Meta's own ecosystem and can seamlessly access Instagram and WhatsApp. David Singleton, who leads the engineering development of personal agents, previously served as Chief Technology Officer at Stripe and CEO of AI startup Dreamer, and joined Meta earlier this year. He mentioned that if the services users rely on have APIs, "you can directly have Muse build custom connectors for you," and the team will continue to expand official partnerships and service integrations. Muse is powered by Meta's latest AI model, Muse Spark 1.3, and will in the future also be integrated into Meta's AI smart glasses. Strategic Positioning: The First Step Toward "Personal Superintelligence" In its launch statement, Meta described personal superintelligence as "one of the most transformative technologies of our lifetime" and positioned Muse as the first step toward realizing that vision. From a market strategy perspective, Meta chose to embed Muse directly into WhatsApp, a messaging platform with a massive user base, in an attempt to achieve product penetration without changing users' habits. In terms of business model, Muse offers both a free version and paid tiers, with paid plans priced at $20 and $100 per month, which provide access to greater computing power. Wang and Singleton expect most users to use the free version, while the paid tiers target users with higher demands. In terms of talent acquisition, Meta's Meta Superintelligence Labs (MSL) continues to aggressively recruit top AI engineers. Last year, Meta brought in Ruoming Pang, a former executive from Apple's AI model division, and recruited several prominent researchers from OpenAI; Singleton also joined this year to lead Muse's engineering development. Meta also announced that Muse will launch a Confidential VM version later this year, further strengthening privacy protection through end-to-end encryption, which is expected to become a key differentiator in the field of AI agent security.
/ Read Next Article /
Source: https://www.163.com/dy/article/L6CC3F8H05198NMR.html If there are any copyright issues, please notify us via email: 13951431913#139.com or submit a comment on the right.