Meta launches Muse, a personal AI agent that can execute tasks like booking travel, shopping and managing email across connected apps

Meta has launched Muse, a consumer AI agent available now in the US via web, iOS/Android and WhatsApp, with AI glasses support planned. Muse can take actions on a user's behalf, such as sending emails, booking travel, filling forms and making purchases (via Link by Stripe, with Shop Pay and 1Password coming), by connecting to apps through built-in connectors, user-supplied API credentials, or browser access. It runs on a new Meta model called Muse Spark inside an isolated "Muse Secure VM," with a separate "Sentinel" agent for safety oversight. Pricing is free, or $20/month (Power) and $100/month (Maximum), and a payment card is required to sign up even for the free tier.

This is a real, shipped product with named infrastructure, specific pricing and stated payment rails, not a demo or roadmap tease. It matters most for distribution: Meta is pushing an action-taking agent through WhatsApp's existing user base, which is a meaningful move against Google's Gemini Spark and Anthropic's Claude Cowork. Developers and payment/commerce players (Stripe, Shopify, 1Password) now have a concrete integration target, and anyone evaluating agent platforms has real pricing and architecture to compare against. That said, there is no usage or adoption data yet, so the actual market impact is unproven.

The framing that this marks the end of the "ChatGPT era" is marketing language, not a demonstrated shift, since no independent performance or reliability data exists yet. The heavy emphasis on Meta's privacy and trust history (Cambridge Analytica, FTC settlements, the $18B multistate and $942M New Mexico judgments) is legitimate context but is doing a lot of the article's engagement work rather than assessing whether Muse itself is secure; the "Secure VM" and "Sentinel" safety claims are Meta's own descriptions and have not been independently verified. Coverage here is secondary, relaying Meta's announcement with no primary source links captured.

1. 01Independent security or red-team assessments of the Muse Secure VM and Sentinel isolation claims, particularly around payment and email access
2. 02Actual adoption numbers or usage disclosures from Meta in the next earnings call or product update, since none exist at launch
3. 03Whether competitors (Google's Gemini Spark, Anthropic's Claude Cowork) announce comparable action-taking capabilities or pricing within the next one to two months, indicating a genuine competitive response rather than a one-off launch