Meta on Tuesday rolled out a long-planned personal artificial intelligence agent aimed at helping users complete daily tasks.
The Muse agent can autonomously send emails, book travel on a person’s behalf, sell a car and lower a bill, according to the social media giant.
Modeled on the open-source AI agent OpenClaw, Muse is designed to access a person’s apps across categories like email, calendar, payments, health, shopping and the smart home, Meta said. People choose which apps it connects to and can revoke access at any time.
OPENAI UNVEILS GPT-6 ASTRA WITH MAJOR ADVANCES IN AI CAPABILITIES: ‘NOW IN THE AGI ERA’
Each Muse agent runs on its own virtual machine, a cloud-based emulation of a personal computer, which enables it to keep carrying out requests in the background even when a person is not actively using it.
Meta said that users can tell Muse what they need to get done, and it will take action using the Muse Spark model that the company built for real-world agentic work. It can handle tasks such as sending an email or booking travel, as well as bigger projects.
NVIDIA CEO JENSEN HUANG DECLARES ‘AGI HAS ARRIVED’ AFTER OPENAI UNVEILS GPT-6 ASTRA
| Ticker | Security | Last | Change | Change % |
|---|---|---|---|---|
| META | META PLATFORMS INC. | 613.48 | -3.29 | -0.53% |
The company said that once a user shares a goal with Muse, it helps them develop a personalized plan and coordinate their time and resources, while advancing the work on its own – opening browsers, filling out forms, and even negotiating on the user’s behalf.
For longer tasks, Muse can continue to work after the app is closed but can come back when something changes or if it needs approval, like before sending an email or making a purchase, Meta said.
Muse also remembers a user’s preferences, giving it the ability to make suggestions unprompted and take action on details that were only mentioned once. As an example, Meta said that Muse can turn a recipe reel saved on Instagram into a grocery list, then generate a suggested menu for a dinner party that takes into account their friends’ dietary restrictions from prior gatherings before sending invites.
AI INFRASTRUCTURE INVESTMENT PROJECTED TO TOP $31T BY 2050
Syncing up the Muse agent with various apps that may contain a person’s real data increases the potential for the new tool, though it also raises the risk of safety and reliability issues for users who granted Muse access to their personal information and if the agent misbehaves.
Vishal Shah, VP of AI products at Meta, said the company initially delayed the release of the product in April to make it more secure. The company determined the extra work allowed it to “cross the threshold” and meet minimum requirements for product safety, security, privacy, model performance and other metrics.
“It is impossible to say that there is never going to be a mistake, but every single part of the architecture has been designed to make this as safe, as secure, as private as we can possibly make it,” Shah said in an interview.
Later this year, Meta is planning to introduce Muse Confidential VM, which will keep a user’s data and communications with Muse in an encrypted format that only the user holds, so that only they can access it.
The product will initially be available only in the U.S. through a dedicated Muse app or Meta’s WhatsApp messaging service, Meta said. The company added that it plans to add Muse to its smart glasses “soon” without further details.
Reuters contributed to this report.
