Runlayer now supports enterprise governance for Muse.
Employees are giving Muse access to their company's most sensitive data & systems. Yet enterprises have no way to govern or monitor this access. Many are blocking it. Others have had Muse go rogue, leaking company data, sending unauthorized emails, and more.
Runlayer solves this by providing secured connectors for Muse, governed by deterministic policies and runtime security models that you can control. Runlayer also detects when Muse is being used within company systems without governance and alerts you.
Runlayer gives security teams peace of mind with visibility into tool calls while letting employees use the app they prefer most.
What is Meta Muse?
Muse is Meta's personal AI agent. It follows the architecture of the hot personal agents like Instinct and Grok Bot where it gives the agent a dedicated VM to work out of, making it more capable for computer use and long running tasks. These tasks include browsing, filling out forms, and making purchases. It can keep working after people close the app, making it perfect for more complex tasks.
How Muse connects to other apps
Muse ships built-in connectors such as Gmail, Google Calendar, Outlook, and Google Docs. Once connected, Muse can take action on behalf of users. This comes with high risk when connected to company systems though. There have already been widespread reports of Muse sharing sensitive PII without consent, sending external emails unauthorized, and more.
For anything else, people can ask Muse to build a custom connector. Handed a remote MCP server URL, Muse builds an MCP client on its VM, connects over streamable HTTP, and saves the integration as a skill. This path is ungoverned and allows risky actions at home or at work.
Your employees want to use the latest and greatest agents at work. With Runlayer, you can let them since security and control are built-in from the start. Our proprietary connector and identity technology allows security teams to securely connect business systems to Muse.
How Runlayer governs Muse
Muse already knows how to talk to a remote MCP server, so employees don't need a dozen unreviewed connectors. They need one: the Runlayer plugin. It's a single endpoint that gives access to every connector and skill your organization has approved, drawn from 18,000+ supported MCPs plus custom connectors for internal systems.
What Security teams get
Every tool call Muse makes through Runlayer passes through the MCP gateway and inherits your controls.
- On-behalf-of authentication. Each employee authenticates as themselves, so Muse acts with that person's permissions and no more. Groups sync from your IDP via SCIM.
- Policies and approvals. You can set fine-grained rules, for example denying Gmail sends to anyone outside your domain. Human approval rules can gate potentially destructive tool calls.
- Audit logs. Every consequential action is logged and exportable to your SIEM.
- Runtime security. Muse reads the open web, so every tool call is scanned for prompt injection, credential exposure and data exfiltration.
For Muse connections that bypass Runlayer entirely, shadow AI discovery shows you who is using Muse and what they are doing for each individual across your organization.
Runlayer works across any model, any tool, and any device, with support for 300+ AI clients. The setup you build for Muse carries over to the next agent your employees adopt.
AngelList, Gusto, Baseten, and PagerDuty are all using Runlayer to put their employees on the golden path to delegating real work to agents. Book a demo and see how Runlayer secures Muse for your employees.
