Set up response using GPT

In addition to the regular and fixed response types according to user settings such as text response, images, tags, buttons... (learn more: Common response types and settings ), users Users can set up responses using GPT so that the Bot can answer customer questions in a customized way based on Prompt - command or Data source.

After entering data into the knowledge base, users can further configure the Bot to recognize customer intent and use specific predefined topics to provide answers, to improve customer satisfaction and accuracy

First, users need to set up a conversation context (learn more: How to set up a scenario )

After you have a specific context, in the left menu bar -> Click on Build Bots -> select Flow Editor (v1.0)

The Flow Editor (v1.0) will display a list of intents that the user has set up

Select the intention you want to set up a response using GPT, then click the Response tab -> select GPT.

At the response setup section, click enable Response by GPT -> The system will display the setup:

• Response by Prompt: The bot will follow the command (prompt) that the user has entered in this section to respond to the customer when meeting the selected intent. For example: with the intent "need consultation by phone call", the user sets up a prompt response and enters the command "ask for the customer's phone number/email information and say the staff will call back as soon as possible". . -> When encountering this intention, the Bot will follow the command the user has set.

• Response with data source: The bot will retrieve data on topics specified by the user to find the most suitable answer and respond to the customer. See the illustration below:

Note: The bot will only retrieve data on topics that the user has selected to provide an answer.

For example: in the "Chatbot Vagent information" intent, the user sets the response with the data source and selects the "Chatbot Vagent" topic. -> When encountering this intention, the Bot will retrieve reference data from the topic "Chatbot Vagent" to find the most suitable response and reply to the customer.