Introduction

The NLP Q&A component in wolkvox Studio allows you to incorporate an Artificial Intelligence bot that answers customer questions based on mandatory training (topics and context defined by your operation). This component is part of the “Cognitive” group and is available for Voice, Interaction, Chat, and CRM + Webhook routing types.

NLP Q&A does not respond to the end user on its own: its result is saved in a variable (by default $nlp_QA), and you must use an output component (for example, Answer chat) to show that response to the customer.

Configuration

1. Add the component to your diagram and open the configuration panel by double-left-clicking.
2. In the configuration panel, you must do the following:
   • Load a template or create a new one: At the top of the panel, you will find the template selector (for example, default) and the “Load template” button.
     • To use an existing one, select it and click “Load template.”
     • To create a new one, type the template name in the corresponding field and execute it with the load template button. The goal is to leave a template active to begin training.
   • Train the bot from the “Configuration” tab.
     • In “Title,” type the name of the topic (for example: “Warranty policies,” “Business hours,” “Order status”).
     • In “Training text,” paste the full content that the bot should learn for that topic.
       • Limit per item: up to 2000 characters (spaces count as characters).
       • Operational recommendation: this field is not compatible with accents/tildes, so the text should be written without them.
     • Click “Add” to save the training.
     • Verify that the item is listed in the table (you will see columns such as Template, Title, and Context).
     • If you need to adjust an item, right-click on the record in the table and use “Edit item” or “Delete item.”
     • Check the “Total template text” indicator, which shows the character consumption of the template against the maximum allowed: 10,000 characters per template.

Test the bot in the “Questions and answers” tab (mandatory). This validation is essential: if you do not test the bot, the system will not allow you to save the component configuration.

• In “Question,” type a real question a customer would ask.
• Click “Test.”
• In “Result,” validate that the response is consistent with the uploaded training.
• Confirm the “Response variable”: the component saves the answer in $nlp_QA (this variable cannot be changed).
• After executing the Test and validating the Result, click “Save” to apply the changes.

Remember: NLP Q&A generates the response but does not send it.

1. Add an output component, for example “Answer chat.”
2. In the output message, insert the variable $nlp_QA so the customer sees the generated text.