Your conversational AI chatbot may process personally identifiable information (PII) while interacting with users.

UI Builder includes AI-powered capabilities that automatically detect and redact sensitive information in user utterances and bot responses. Examples of PII that the AI can redact include, but are not limited to: names, job titles, phone numbers, email addresses, postal addresses, and more.

<aside> 💡

Explore more examples of the types of PII your conversational AI can redact here.

</aside>

Supported Levels of PII Redaction:

  1. Allow All PII Recording – No filtering; all user input is logged in backend analytics.
  2. Filter PII Recording – PII is redacted in backend analytics while preserving the rest of the utterance.
  3. Block All PII Recording – No utterances are logged in backend analytics.
  4. PII Filtering on User Input - PII is redacted in real-time

These options help ensure compliance with privacy regulations and internal policies. You can configure PII redaction directly within UI Builder's designer canvas, just like any other activity.

image.png

As an example, below is a conversation workflow that begins with a PII activity. The user is prompted for some information and that data is stored in a variable. Then, the contents of the variable are displayed to the user. Here’s what the conversation design looks like, starting with the Allow all PII recording option:

image.png

Using the WebChat simulator, the conversation is run three times - first with the Allow all PII recording activity, then with the Filter PII recording activity, and then with the Block all PII recording activity. Each time the conversation is run, the user enters, “My name is John and my phone number is 555.1212.” and the chatbot displays the entry back to the user.