Why context matters in private data redaction

Every day, your team records customer interactions to make sure you're acting in compliance with federal, local, and organizational guidelines.

When a payment is collected, the consumer might share their payment card information (PCI) and their personally identifiable information (PII). 

And since you're recording that interaction, handling the data safely and securely is critical to your reputation and your bottom line (not to mention staying in compliance with CFPB regulations).

That means redacting data from the recordings, but doing so accurately and efficiently has been a challenge. 

Meet a more accurate redaction model

Reliability and accuracy are the most important qualities in a redaction partner. Prodigal has built a state-of-the-art Redaction AI model to handle PCI and PII data.

Given our focus on the consumer finance vertical and our Prodigal AI Intent Engine design, we handily beat out Amazon’s redaction models in accuracy.

Redaction experiments and examples

Here's a (fictitious) sample call where you can see how Prodigal's redaction works.

We've highlighted the pieces of information our AI redacts, and tagged it with the category that requires redaction. The numbers you see after those tags are our model's confidence in its accuracy - 1.0 represents total confidence.

Those are pretty high confidence numbers, and as you can see, Prodigal's model is catching all that personal information.

But what if the caller is difficult to understand, or something else causes an error in transcription?

Here's that same call, with some changes made to show you what happens:

social → vocal

phone number missed

card → car

CVV code missed

As you can see, those changes cause a decrease in the model's confidence level, but it still correctly identifies the personal information we need to redact.

What that shows us is that you can't just rely on old-fashioned transcription models. You need redaction that understands context.

That's why we've focused on consumer finance. 

Our AI Intent Engine is trained on over 300 million consumer finance interactions, so it doesn't just rely on transcriptions, but truly understands the context of the conversation.

The importance of solid redaction models

As you can see, context and accuracy count when it comes to compliance. Prodigal’s redaction model offers: 

  • Better accuracy than Amazon's model
  • More complete coverage on PII
  • Contextual understanding to deliver the best results

The bottom line: You need a redaction partner you can trust so you know you're being compliant and taking care of your customers. Look for accuracy and context to deliver that confidence.

"We found Prodigal while looking for solutions to reinforce our call recording safeguards and further protect our customers’ personal information.
We looked at multiple vendors but ultimately trusted Prodigal and their industry-trained AI models to get the job done. They were great to work with and further customized their outputs to meet our specific needs. We’d recommend them to any team looking to effectively protect their consumer data and strengthen compliance." -VP of InfoSec, Policygenius