To make your smrtDialer campaigns as efficient as possible, especially when you are using multi-line dialing, we employ artificial intelligence (A.I.) to detect whether a human being answers the call or it is a recorded message.

How to turn AI on for smrtDialer Campaigns

Whether or not AI is used in a campaign is controlled by the agent. Start a New Call Session by clicking on the green button. Once the campaign is ready to run, you can select from the ⚛️ Settings menu at the top whether to use AI Voice Detection or not.

How does AI work?

Once a phone call is answered, the technology listens for evidence that it is a live human or a recording. The "evidence" is based on millions of calls placed through our underlying carrier, that are continually studied to sharpen the AI detection skills. Typically, humans answer quickly after the ringing ends and speak in short introductions. Machines tend to have a delay before answering and speak in longer, more fluid introductions.

In smrtDialer, the default is that it will listen for up to 30 seconds to determine whether it is a human or a machine on the other line. However, it rarely takes that long to make its determination.

Using AI Detection for Voicemail Drops

If the AI detects that it is not a live person answering, it assumes that it is a voicemail. It will listen for the beep and leave your voicemail recording.

You may notice that when it detects a voicemail, the line still stays active for the full 30 seconds of detection. Do not worry! Your voicemail recording will not be cut off or have a long stretch of blank noise/dead air. Once the 30 seconds are finished, the line will free up for the next call.

Troubleshooting voicemail drops >>

Limitations of AI in smrtDialer

As great as the artificial intelligence is in smrtDialer campaigns, it is not infallible.

There are some limitations to the technology, which is why we allow you to control whether it is active or not each time you start a call session.

  • It may mix up if it is a human or a machine. For example, sometimes the outgoing greeting of a lead is recorded in such a way that it sounds like a human answering the phone. Or, the human may pick up but not speak immediately, so the AI thinks it is a machine.

  • There may be a couple seconds delay, as it works to understand if the call has been answered by a human. Most often this manifests as the call recipient saying "hello?" twice.

  • If the outgoing voicemail greeting of your lead is longer than 30 seconds, your voicemail recording may not be left in full. For example, if your voicemail recording is 10 seconds, but the outgoing greeting is 35 seconds, only the last 5 seconds of your voicemail recording will be heard.

Did this answer your question?