AI

Setting AI guardrails

How to configure what the AI can and can't do on your behalf.

Guardrails are the limits you set on what the AI is allowed to do autonomously. Ship them before you ship the AI live.

Autonomy levels

  1. Observation only — the AI drafts a reply; you send it
  2. Draft + review — the AI sends after a 3-minute hold; you can cancel
  3. Auto-send with exceptions — the AI sends immediately unless the lead matches an exception rule
  4. Full auto — the AI sends without review for any matching trigger

We recommend starting at level 2 for the first week, watching the drafts, and moving up once you trust the output.

Common exception rules

  • Leads on listings over $2M — escalate to you
  • Any message containing 'legal', 'lawyer', or 'dispute' — hard stop, escalate
  • Leads outside your licensing region — escalate and optionally refer
  • Any reply to an offer or counter-offer — always human

Ready to set it up?

Start a free trial and we'll walk you through this during your setup call.

Start free trial