AI Interaction Data Capture is a way for Talkative to automatically classify and assign interaction data values.

Interaction Data is metadata to conversations, with user-defined key-value pairs. Typical Interaction Data values include "Name" and "Email", but can also include "Category", "Account Number", or anything else you desire.

Instructions to use AI Interaction Data Capture

This feature is available on all Talkative subscriptions. You must be an Account Holder to configure it.

Step 1: Access the template list view

• Search for "Capture" in settings

Step 2: Create a config

• Create a new configuration
• You can create multiple configurations and map them to different queues
• You can only map one configuration to a specific queue
• There are no limit to the number of configs

Step 3: Create values and assign to queues

• Choose the queue(s) you want the config to work for
• Create Interaction Data fields
• Think of the Interaction Data description as a prompt you are giving to the AI to categorise the data based on the transcript

Step 4: Watch it work

• Agents can view the Interaction Data in the console under the information tab
• You can view the data collected in the Interaction Log page
• The system updates on every new message
• The system also works with video chat interactions

Note: This feature involves sending your transcripts to OpenAI to perform analysis and data classification

Tips

Here are some data fields you might want to try out:

• Customer Status: New vs. Returning Customer, Loyalty Level (e.g., Silver, Gold, VIP).

• Customer Satisfaction/Outcome: Resolved, Unresolved, Escalated, Follow-up Needed, Abandoned.

• Language Used

• Product/Service Mentioned

• Request Type: Information Request, Issue Resolution, Order Placement, Product Feedback

• Follow up required?

• Complaint yes/no?

You can use Analytics custom reports and interaction logs search filters to retrieve interactions with certain AI data types required. 

You can apply Post-Interaction Wrap Up Forms to allow agents to check/override Interaction Data at the end of the conversation as part of their wrap process.