Security and control
The platform is designed so an AI employee only does what you enabled, with your information isolated from other companies and every action recorded.
Read the Privacy Policy →
Every request goes through these layers
Your organization
Isolated data
The employee
Role and assigned training
Registered skill
Declared read or write
Enabled connection
Encrypted credentials
Human approval
When there is a write
And everything stays in the activity log, with its result and its cost.
Every employee, piece of knowledge, connection, task and conversation belongs to an organization. Queries, actions and downloads are filtered by organization, and another organization's resource responds as if it did not exist.
Your connections' tokens and keys are stored encrypted, never shown in the interface or written to logs.
An employee only uses the connections you enable. Whoever delegates work does not inherit another employee's tools or connections.
Every external connection requires HTTPS, validates the domain, blocks private networks, pins the resolved address, does not follow redirects and has time and size limits.
Employees use declared and reviewed capabilities. They never run arbitrary code or instructions requested by a model, a user or a document.
Actions that change external systems, such as creating an event, publishing a pull request or calling a write API, wait for a person's approval.
Training and external data are presented as reference information and never replace the employee's rules or permissions.
Every job records activity, the tool used, the result and the usage, along with the person who requested or approved it.
Secure session with email verification, two-factor authentication and passkeys. Public forms have usage limits.
Temporary files are deleted automatically, so they stay available as long as needed to work and no longer.
Create your account, receive USD 2 to hire your first employee, and request a promotional USD 1 to try it.