AI voice agent

Guardrails, and a clean hand-off when they're hit

Say what the agent may discuss, what it must disclose, and what it should never attempt. When a call goes past the line, it escalates to a human instead of improvising.

Sales and Enterprise plans

What it does

Rules the agent actually runs on

What you set on the guardrails page is what drives the live agent — not a document it might have read.

Escalation as a tool

Hand-off is an explicit action the agent can take, with the wording decided server-side rather than improvised.

Auditable

Escalations and blocked topics show up in call analytics, so you can see what keeps hitting the wall.

How it works

01

Define the boundary

Allowed topics, forbidden topics, required disclosures.

02

Set the hand-off

Who takes the call, and what the agent says when it hands over.

03

Watch the log

Analytics shows which calls escalated and why, so the boundary can be tightened or relaxed.

In the box

Everything guardrails & escalation includes today.

  • Allowed and forbidden topic lists
  • Required disclosures
  • Escalate-to-human tool
  • Server-decided hand-over wording
  • Escalation reasons recorded per call
  • Applies to the web widget and phone calls
  • Shared across an organization's agents

Stop scheduling. Start closing.

Everything above runs in one product, on one plan, from one dashboard.

7-day Pro trial, no card required. Set up in under three minutes.