Trust center

The agent acts on your behalf — never behind your back. Every action passes through consent, lands in an audit log, and can be undone. And your data stays protected by encryption, isolation and 152-FZ compliance.

Agent trust

The agent acts in the open

When the assistant runs a task, it doesn't get a blank cheque. Data changes and outbound calls pass through explicit control — you see the action, approve it, and can roll it back.

Consent for actions

Before changing data or sending anything to an external service, the agent asks for permission. Outbound data through integrations never clears automatically.

A log of every action

Every executed action is written to an append-only log — what, when and with what result. Zero silent writes: everything the agent did is visible.

Reversible changes

The agent prepares a rollback plan for every change. If something goes wrong, the action can be undone and the previous state restored.

Autonomy levels

You choose how independent the assistant is: read-only, or automatic execution of safe edits. Switch the level in one click, any time.

Principles

We never use your data to train models. Access belongs only to you and the people you grant it to. Security isn't a feature — it's the foundation of the architecture.

How we protect your data

Encryption

Data is transmitted over secure channels (TLS) and stored encrypted. Access keys are isolated and never leave the secure perimeter.

Tenant isolation

Every organization's data is logically isolated. One customer never sees another's data — at both the database and application level.

FZ-152 compliance

Personal data is processed and stored on servers in Russia certified to FZ-152 requirements.

Backups

Regular automated backups with point-in-time recovery. Your data isn't lost when failures happen.

Compliance

The platform complies with Federal Law No. 152-FZ "On Personal Data". Data is processed and stored within Russia. We run regular security audits and never share data with third parties without your consent.

FAQ

What people usually ask about security

Still have questions?

We'll walk you through how the platform's security works

Contact support