Skip to main content
Audact
See the offer

Integration · Cal.com

The agent books against your real availability.

Integration · Cal.com: In controlled testing

In controlled testing: Cal.com event types would become tools the agent may call. It would check live availability against the rules on the event type (working hours, buffer, round-robin host assignment), propose two or three slots in the caller's timezone, confirm, and Cal.com would send the invite and the reminders. No human escalation for a booking.

The tool the agent calls

{
  "tool": "cal_book_slot",
  "arguments": {
    "event_type": "intake-30m",
    "start": "2026-09-03T09:30:00+02:00",
    "timezone": "Europe/Amsterdam",
    "attendee": { "name": "M. de Vries", "email": "m@example.nl" }
  }
}

Worked example. Acme GmbH and Nova Industries are fictional; no customer data appears here.

Availability is read at the moment of the call, not from a cached copy.

What it does

Would read live availability per event type, including buffers and round-robin assignment.
In controlled testing
Would book, reschedule and cancel from inside the conversation.
In controlled testing
Cal.com would send the invite and the reminders; Audact would record that the booking happened.
In controlled testing

Setup

Planned setup

In controlled testing

  1. Connect Cal.com from your Audact tenant.
  2. Select the event types the agent is allowed to book against.
  3. Configure which caller fields are collected: name, email, phone.

Questions

What happens if the slot is taken between proposal and confirmation?
In controlled testing: the booking would fail and the agent would propose again. The race is real and it would be handled in the conversation rather than in an apology email afterwards.
Does the disclosure happen before the booking?
In controlled testing: yes. Article 50 would ask for the information at the first interaction, which is well before anyone gets to a calendar.

What the caller hears first

Voice shows how the disclosure, the policy and the booking sit in one conversation, in the order the law asks for.

See Governed Voice AI

Related