Meta debuts Muse AI agent to automate everyday tasks

Meta has introduced Muse, a new personal AI agent designed to automate everyday consumer workflows in the United States. Powered by the company's Muse Spark model, the agent integrates directly with user applications like email, calendars, and smart home systems. Muse can perform complex actions on behalf of users, such as booking travel, lowering bills, filling out forms, and making purchases. To facilitate transactions, the agent leverages Link by Stripe for secure checkout, with integrations for Shopify's Shop Pay and 1Password scheduled to launch soon. The service is initially available via muse.ai, iOS and Android apps, and WhatsApp, with future integration planned for Meta's AI glasses.

While Muse is free to use, it requires a payment card to get started and features a usage meter to track consumption. For heavy users, Meta offers two subscription tiers: the Power plan at $20 per month and the Maximum plan at $100 per month. To address privacy concerns, Meta runs the agent inside a dedicated virtual machine called Muse Secure VM, which isolates the AI in its own secure browser environment. A separate Sentinel agent runs on the same virtual machine at the system level to prevent Muse from accessing passwords or payment methods, and Meta claims conversation data will not be shared with its advertising systems.

This high-access architecture enters a market where competitors like Gemini Spark, Claude Cowork, and Instinct are also vying for agentic control. However, Meta faces a steep hill regarding consumer trust. The launch follows a massive $18 billion multistate settlement over social media harms and a $942 million settlement in New Mexico. Meta's history also includes a $5 billion FTC settlement in 2019, a 2011 FTC privacy settlement, and a 2019 incident where user passwords were exposed in readable formats.

For AI developers and industry practitioners, Muse represents a major shift from passive chatbots to active, transaction-capable agents. Meta's approach of allowing users to connect services via public APIs when built-in connectors are unavailable establishes a new paradigm for open-ended agent integration. If Meta can overcome its historical trust deficit, the deployment of Muse Spark could accelerate the mainstream adoption of autonomous consumer agents, forcing competitors to match its secure virtual machine architecture and transparent opt-in controls.