Security

Trust, built in from day one

We tell you exactly what's live and what's coming. No certifications are claimed during the preview.

Our safeguards

Per-user data isolation

Planned

Every record is locked to its owner and workspace with database-level rules.

Server-side AI & SMS limits

Planned

Hard daily caps and spending budgets checked on the server before any call.

No runaway loops

Planned

Only a new human message can trigger KAI. KAI never replies to itself or to delivery receipts.

Human approval & Pause KAI

Demo only

Sensitive topics always come to you. One tap stops all automation.

AI activity audit log

Demo only

Who acted, in which mode, when and why.

Encryption

Planned

Encrypted in transit and at rest. KAI is not end-to-end encrypted, because the AI must read allowed messages.

Block & report

Demo only

Stop unwanted contact and flag abuse.