Meta has launched Muse, its first consumer AI agent, built to take on tasks that usually require several apps and a fair amount of attention. Muse can browse the web, shop online, fill out forms, plan trips, send emails, schedule events and handle much more.
The main idea is simple: tell Muse what you want done, and it can manage the steps needed to reach that goal. Meta says, “Once a person shares a goal with Muse, it helps them develop a personalized plan and coordinate their time and resources, then advances the work on its own.”
Muse was designed for people with “no technical experience” at all. Meta also says the agent is more accessible than agents from other platforms, with a focus on completing complex, multi-step tasks without requiring users to guide every move.
How to access and use Muse
Muse is separate from the existing Meta AI app. It has its own iOS and Android apps, a website at muse.ai, and access through WhatsApp. The agent operates within a secure virtual machine, whether someone uses the standalone application or WhatsApp.
Users do not need a Facebook or Instagram account to use Muse, but they do need a Meta account. The service is available at first to people aged 18 and over in the US, with Meta planning a rollout across the US on iOS, Android and muse.ai. AI glasses are “coming soon.”
Muse can connect with third-party services such as Google and Spotify. A service can connect to the agent if it has a public API, giving Muse a way to work across more of the tools people already use.
Shopping requires one extra setup step. Users must link a credit card, with that payment connection facilitated by Stripe, before Muse can make purchases.
Approval checks and privacy protections
Muse is meant to handle work without constant supervision, but it does not get unlimited control. Meta says the agent will check in for approval before actions such as sending an email or completing a purchase. That gives users a chance to review important steps before they happen.
Meta has also built safeguards to stop Muse from going rogue and to protect user data. Each person’s information stays separate inside a dedicated virtual machine. Meta plans to introduce a more secure “confidential” version of that virtual machine later this year.
There is still a clear warning attached to the launch: “Muse can and will make some mistakes,” Meta says. That matters because the agent is designed to work across several steps, make plans and continue tasks after a user leaves the app.
Meta says Muse can keep working in the background after users close the app. It can return if something changes or if it needs approval, which makes the service less like a chatbot waiting for a reply and more like a task assistant working toward a goal.
Privacy will be a major part of how people judge Muse. Meta has faced privacy backlash and privacy-related ordeals in the past, including a figure stating that Meta would pay up to $17bn in a huge child protection settlement. Against that backdrop, users will want clear boundaries around what Muse can access and how their conversations are handled.
Meta allows users to opt out of having their interactions used for AI training. The company also says it will not share details from conversations with Muse with its advertising systems, though some interactions could influence ads indirectly.
A new piece of Meta’s AI strategy
Muse runs on Meta’s in-house model, Muse Spark. Meta says the agent can work proactively, help people get things done and develop personalized plans instead of only responding to individual prompts.
Vishal Shah, Meta’s VP of AI Products, described the goal as helping the system “build its own software.” Meta calls Muse “a first step: an agent that takes on more of the work so people can focus on what matters to them.”
The launch also fits into a much bigger business push. Meta has spent billions overhauling its AI strategy and wants to catch up with rivals such as OpenAI, Anthropic and Google. Muse gives that strategy a direct consumer product, where the test is not just whether an AI model can answer questions, but whether it can complete useful tasks across the web.
Most of what people need from Muse will be free, while Meta plans paid subscriptions for additional features. For now, the service’s appeal rests on a practical promise: users can describe a goal, let Muse organize the work, and step in when the agent asks for permission.
That promise comes with limits. Muse can make mistakes, purchases need a linked card, and sensitive actions still require approval. But Meta is positioning the agent as a way to hand over more everyday work without requiring technical knowledge or a Facebook or Instagram account.
Based on
- How to get started with Meta’s new AI agent, Muse — engadget.com
- Meta’s Muse AI works and creeps me out | The Verge — theverge.com
- Meta Reveals Its AI Agent That Can Shop, Send Emails And Plan Trips On Your Behalf — engadget.com
- Meta bets on AI agent Muse to catch up in AI race | The Verge — theverge.com
- Muse: Meta launches new AI agent – here’s what to know | The Independent — independent.co.uk
