Skip to content

Feature · pacing

Predictive dialer: pacing math, dropped calls and the 3% cap.

PyAI has no pacing engine: it places one call per request, and an AI agent answers every pickup.

Get a sandbox key$50 credit. No credit card required.
No overdial ratioAgent present on each answerAMD $0.004 a call3% cap explained

What PyAI gives you for predictive dialing

  • A predictive dialer rings several numbers per free agent from a forecast of answer rate and handle time, so at a 25% answer rate about four lines ring per agent, while PyAI places one call per request with an Omni agent already on the line.
  • US rules in 47 CFR 64.1200(a)(7) cap abandoned telemarketing calls at 3% of calls answered by a person, measured per campaign over any 30 day window.
  • PyAI runs no pacing ratio. Each request rings exactly one destination at $0.05 per minute plus $0.01 telephony, with no seat license and no platform fee to grow with headcount.
  • Answering machine detection costs $0.004 a call and returns who answered, the kind of answer and a confidence score, so your dialer screens voicemail and fax without tuning an overdial ratio.

What does the person hear when a pacer guesses wrong?

Compare an overdialed answer with an AI answered one. The lines are an example script, not a recording, drawn from the outbound call and AMD flows.

Pick a moment, then press play

Person calledAI voice agentDialer or PyAIHuman agent
Ready00:00

What happens on an AI agent call that a pacer would overdial?

A predictive dialer guesses when an agent will be free. On PyAI, every answered call gets its own AI agent. How many calls run at once depends on how many requests your system sends.

Request, not ratio

Your system asks for one call to one number. Concurrency is how many requests you send, within your account's concurrency limits.

Checks before ringing

PyAI evaluates the outbound policy, consent, suppression, quiet hours and a per-destination velocity cap.

Answer meets an agent

The telephony edge opens an outbound session for the agent the instant the person picks up.

Status flows back

Ringing, answered and completed or failed reach your system, along with the recording and transcript.

Machine detected

On PyAI numbers with managed AMD on, voicemail, fax and invalid numbers are hung up automatically.

Rules block the number

A compliance block (403) returns before any ring, so the attempt never reaches the person.

When does a call center still need a predictive dialer?

Pacing still helps when people must talk and the list is huge. AI calls fit when the script is short and the answer is the same each time.

Pacer stays

Big cold lists

A hypothetical 40-seat floor working 20,000 cold numbers a day at a 10% answer rate gains real talk time from overdialing, as long as abandonment stays under 3%.

Humans required

Licensed conversations

Insurance or loan calls that need a licensed person on every connect still require human agents, so a pacer keeps them busy.

Move to AI

Reminders and confirmations

Appointment confirmations with consent on file run as single AI calls. Nobody waits, and the agent reschedules through a server tool.

Mix both

Screen, then hand off

An AI agent qualifies first and transfers interested people to a human, so the floor only spends time on warm conversations.

Hand it to AI

Payment reminders

A utility calling customers who opted into balance reminders can let the agent state the amount due and transfer anyone who disputes it to a billing specialist, with the whisper explaining the dispute. Recap at $0.02 a minute then logs the promised payment date for the collections team.

Pacer stays

Political or survey floors

High volume research floors with trained interviewers keep pacing, because the interviewer, not a script, is what the client pays for.

Predictive dialing questions buyers ask

Are predictive dialers legal?

Predictive dialing is legal in the US with limits. 47 CFR 64.1200(a)(7) caps abandoned telemarketing calls at 3% of calls a person answers, per campaign over 30 days, and artificial or prerecorded voice calls need consent under 47 U.S.C. 227(b). None of this is legal advice.

How do you choose predictive dialer software?

The right option is the one whose pacing keeps abandonment under 3% on your answer rates, reports it per campaign, and integrates with your CRM and agent desktop. For short repeatable calls, PyAI AI voice agents remove the need to pace human agents at all. Ask each vendor for a sample abandonment report before signing.

How much does a predictive dialer cost?

Pricing for these systems usually combines a monthly seat license per human agent with per minute calling charges, and sometimes a separate fee for each phone number. PyAI charges no seat or platform fee: Omni calls cost $0.05/min plus $0.01 telephony, and AMD adds $0.004 a call when used.

When is a predictive dialer still the better choice?

A predictive dialer still suits a floor of licensed people working a cold list, where pacing keeps reps talking instead of waiting through rings and voicemail. PyAI suits calls that follow a short script, because an Omni voice agent handles every answer at $0.05/min plus $0.01 telephony with no pacing ratio to tune.

PyAI is voice infrastructure. It hands you the audio and transcript, and does not replace your own models or review tools.

Ring your own phone and hear an agent speak the moment you answer.

Start building with $50 creditSandbox keys never bill, and you can start without a credit card.

One stack: Omni, Hear, Speak, Recap and Trace.