A voice that fits the moment

Why we built it

Collections calls are rarely neutral. A consumer may be explaining a job loss, a medical bill or a charge they dispute, and how the agent sounds in that moment decides whether they keep talking.

proAgent's earlier voice model read every line the same way. A hardship explanation got the same delivery as a balance confirmation, with no pauses and no change in pace. While the agent looked up an account or processed a payment, the line went quiet. Consumers notice that flatness, and some hang up before the conversation reaches a resolution.

How it works

proAgent now runs on Eleven v4 Turbo, the top-ranked AI voice model on Artificial Analysis' leaderboard, with voice controls Prodigal built on top of it.

• Each line the agent speaks can carry cues for tone, such as calm or warmth, along with pauses, emphasis and breaths.
• The cues are written once and work across voice providers.

The new voice responds as quickly as the previous model and costs the same per character.

What you get

When the agent sounds like it is listening, consumers stay on the call. More right-party conversations reach a payment or a plan, and fewer are passed to your team after verification. A consumer going through a hard moment hears a voice that fits it, which makes the conversation easier for them and more productive for you.

Talk to your Prodigal team member to enable it on your voice agent.