← Back to the blog

How do you automate appointment reminders with voice AI?

Clay diorama of Peach, the PeachDesk mascot, in a workshop where conveyor belts carry glowing message envelopes

You automate appointment reminders by connecting three pieces: a trigger that starts the call when an appointment approaches, a reminder script that confirms or declines the booking, and a write-back that updates your system of record. In PeachDesk the trigger is an API trigger node or a CSV campaign, the script is a voice agent flow with confirmation capture and a rebook path, and the write-back is a signed webhook or an HTTP request to your scheduling system.

This guide walks through the call-flow design step by step, then works through the per-reminder cost on each sourcing mode. For the scheduling end-to-end picture, see the AI appointment scheduling use case, and for rates see PeachDesk pricing.

How does the reminder call flow work?

A reminder flow has four stages. Each maps to a concrete PeachDesk building block, so the whole design is configuration rather than custom telephony code.

  1. Trigger the call. Your scheduling system starts the reminder when an appointment enters the reminder window by calling a public trigger endpoint on the flow's API trigger node. Alternatively, export tomorrow's appointments and run them as an outbound CSV campaign with per-contact personalization, restricted to calling windows by day, time, and timezone.
  2. Run the reminder script. The agent states the business name, the service, and the appointment time, then asks the contact to confirm. The spoken answer is captured as the call outcome: confirmed, declined, or undecided.
  3. Offer a rebook path on decline. On a decline the agent offers to reschedule instead of hanging up. It can check live availability through a custom HTTP tool against your scheduling system, or transfer the call to a human coordinator, with the answer outcome reported back into the conversation.
  4. Write the result back. The flow sends a templated HTTP request to your scheduling system the moment the outcome is known, or your endpoint subscribes to PeachDesk call events and receives an HMAC-signed webhook. Delivery is best effort with up to three attempts, and you can test webhook delivery before going live.

Calls that do not connect are handled by campaign retry rules: busy, no-answer, and voicemail each get their own retry controls, and a finished campaign can redial its non-connections in one action. The telephony leg itself runs on your registered numbers through any of the supported carriers; see PeachDesk telephony.

What does each automated reminder cost?

The math is per reminder call, and it depends on which sourcing mode serves the pipeline. Take a two-minute reminder call as a worked example. On the Bibha Plan at $0.06 per minute, that reminder costs $0.12 all in; at volume, where the rate goes as low as $0.03 per minute, it costs $0.06. On the $59 per month Pay As You Go plan you pay the metered rate against your balance. Compare plans on PeachDesk pricing.

  • Bibha Plan (Frontier mode). PeachDesk-managed providers serve every stage. You pay the per-minute rate and nothing else; there is no infrastructure to run.
  • BYOK. Your own provider keys serve the stage. You pay your speech and model providers directly at your negotiated rates, and calls on your own keys never consume PeachDesk credits.
  • Local mode. A full open-model stack on your own GPUs serves the call. There is no per-minute provider charge for locally served inference; you pay for GPU capacity, which converts the per-reminder variable cost into fixed infrastructure cost at steady volume.

One cost never disappears: telephony stays metered by the carrier in every configuration, including fully self-hosted ones. No sourcing mode makes reminder calls free.

What happens when a contact does not answer?

Unanswered reminders are normal, so the flow treats them as a retry problem rather than a failure. CSV campaigns retry busy, no-answer, and voicemail outcomes with per-reason controls, and campaigns pause themselves when failure rates spike so a bad contact list cannot burn through your budget unnoticed. You can track progress live, filter every call, and export CSV reports with dispositions, durations, recordings, and transcripts for your records.

Put your reminders on autopilot

Connect your scheduling system to an API trigger node, record the reminder script, and let the agent confirm, rebook, and write back while your front desk does something else.

Can an AI reminder call reschedule, not just confirm?

Yes. When a contact declines, the agent moves to a rebook path instead of ending the call. It can check availability in your scheduling system through a custom HTTP tool and offer open slots, or transfer the call to a human coordinator with the answer outcome reported back into the conversation. The final result, confirmed, rescheduled, or declined, is captured as the call outcome and written back to your system.

How do reminder results get back into my scheduling system?

Two ways. You can send a templated HTTP request from inside the workflow at the moment the outcome is known, or you can subscribe your own endpoint to PeachDesk call events and receive an HMAC-signed webhook after each call. Delivery is best effort with up to three attempts, and you can test webhook delivery from the dashboard before going live, so verify the integration against a sandbox first.

Is it cheaper to run reminders on my own keys or my own GPUs?

Both remove per-minute platform charges, differently. With BYOK you pay your speech and model providers directly at your negotiated rates, and calls on your own keys never consume PeachDesk credits. With Local mode a full open-model stack on your own GPUs serves the call, so there is no per-minute provider charge for locally served inference; you pay for GPU capacity instead. Telephony stays metered by the carrier in every configuration.

Talk to an expert

Tell us about your calls and we will come back with a straight answer on fit, sourcing, and deployment. Your message goes to the team at sales@bibha.ai.

Start free

Tell us where to reach you and what you are building, and we will set up your workspace access. Your message goes to the team at sales@bibha.ai.