Security & governance

Powerful automation, without losing control.

KAI is built so that agents can do real work while you keep full oversight of your data and every decision they make.

The six pillars

Trust is the default, not an upgrade.

Tenant isolation

Every customer’s data is logically isolated and access-scoped. Agents only ever see what their tenant explicitly allows.

Human in the loop

Agents propose; people approve. No email is sent, no record changed and no payment moved without a human decision.

Full audit trail

Every prompt, tool call, result and approval is logged immutably — exportable for internal and external audit.

EU AI Act aligned

Transparency, human oversight and risk controls are built in, in line with the EU AI Act’s requirements.

Read-only by default

Connectors begin read-only. Write scopes are granted explicitly, per agent and per system, and can be revoked anytime.

Least-privilege access

Agents receive the minimum permissions needed for the task — nothing broader, scoped to the data they actually touch.

Our promises

What we will — and won’t — do.

Your data stays yours

We don’t train shared models on your business or family data. It’s used to serve you, and only you.

Explainable by design

Every action comes with the reasoning and the source data behind it, so you can trust what you approve.

Automate with confidence.

See exactly how KAI keeps you in control — book a security-focused walkthrough.