Meta's Muse gives every user a cloud computer. Here's how to do the same for your customers

Description and use

Each user receives an isolated virtual machine with a browser and secure vault, enabling safe execution of a personal AI agent.

Problem

Running a personal AI agent for each customer requires isolated, persistent computing resources that can securely store credentials and allow audit.

Outcome

Customers obtain isolated, persistent cloud computers that securely store credentials and allow audit, mirroring Meta’s Muse architecture without building custom VM infrastructure.

Workflow

• 1. Create or resume a user machine via Burrowbox REST API, tagging it with the user’s externalId and enabling a full Chromium browser with persistent logins.
• 2. Store the user’s login credentials in the machine’s vault using the Burrowbox vault API, ensuring passwords never leave the machine.
• 3. Connect the Vercel AI SDK’s MCP client to the machine’s MCP endpoint with the machine‑scoped token, so the agent can run commands on that machine only.
• 4. Execute agent actions (e.g., browser_login) through the MCP client, which interacts with the browser without exposing credentials to the model.
• 5. The agent’s actions are logged and auditable, and the machine remains persistent across restarts.

Evidence

Each user's agent runs on its own virtual machine in Meta's cloud, the "Muse Secure VM".

Source

Burrowbox gives each customer a persistent Linux machine with a browser that stays signed in, an encrypted vault and its own MCP endpoint.

Source

The AI SDK's MCP client connects to the machine's MCP endpoint with the machine-scoped token, so this agent can't reach any other machine.

Source