Smart Predict uses AI to identify if certain, specific criteria were met in a conversation, completely automating the escalation of issues that require review, allowing you to scale your Evaluation throughput significantly.

Smart Predict answers Evaluation questions using AI to predict the correct answer.

Authenticx can AI score your Evaluations with Smart Predict based on results for ML Models, Classifiers, or Metadata. Users can manually override the Smart Predict answers as needed.

Smart Predict is how Authenticx helps you score Evaluations at scale. By using AI to score a large volume of conversations, you gain insight into why customers are calling and how to create efficiencies in your business.

Your Authenticx team can configure any of your modules to use Smart Predict to score Evaluations.

Smart Predict can score questions using several methods:

In an Insights module, for example, you might have the question “Was an Eddy Effect present?”. Authenticx could configure this question to Smart Predict Yes if an Eddy was found by the Eddy Effect model, and No if it wasn’t.
​
You can also use the HIPPA compliance model, the Safety Event models, and more.
​

You’re likely using Classifiers to identify agent skills, topics, or processes.
​
Assigning the classifier results is a matter of telling the question how to be answered if the classifier is found to be TRUE (the language specified in the Classifier was found) or FALSE (the language was not found).
​
For example, on a Quality module we might want "First Impression" = TRUE to score the question as Meets. We’d then want "First Impression" = FALSE to score the question as Does Not Meet.
​

We can set up your modules to use Smart Predict to fill in questions using metadata.
​
For example, calls from California might require a different set of questions than calls from New York. We can configure a question to answer with the caller’s state or any other metadata we’re receiving from you, triggering a cascade of the appropriate questions based on their location.