The launch of business logic for Mr. Wizard #
- Starting from this version, the ‘Mr. Wizard – wolkvox Studio – Business Logic’ option is available.
- To train the Q&A section of Mr. Wizard, you must use a routing point of type ‘CRM + Webhook + Cron + AutoQAi + Mr. Wizard,’ which is pointed from this business logic configuration option.
- Important: When pointing to the desired routing point, the training that will work to respond to questions asked by the agent to Mr. Wizard will be the one built through a flow in the pointed routing point. In other words, for the Q&A section, what’s trained in the ‘Training’ tab will no longer be taken into account.
A series of new variables has been incorporated for Mr. Wizard in order to create routing based on the operational center’s needs and to tailor the training given to Mr. Wizard for obtaining specific and personalized responses.
- $channel_MrWizard: Indicates the communication channel being used for the interaction with the customer. It can be “voice” or “chat.”
- $total_conversation_voc_rt: Contains the entire conversation of the customer during the interaction. Stores the textual record of all interactions and responses between the customer and the system or agent in real time.
- $last_phrase_voc_rt: Stores the last sentence or phrase spoken by the customer during the conversation. This variable is updated in real time as the customer speaks.
- $idagent_voc_rt: Stores the identification of the agent handling the interaction with the customer.
- $idinteraction_voc_rt: Is the unique identifier for the ongoing interaction. Allows tracking and labeling of each conversation.
- $ani_voc_rt: If applicable to the channel, stores the source number of the interaction, i.e., the number from which the customer is calling.
- $language_voc_rt: Indicates the language being used in the conversation, such as “es” for Spanish or “en” for English.
- $duration_voc_rt: Displays the duration of the interaction in seconds, from the start to the present moment.
- $calltype_voc_rt: Describes the type of call being conducted. It can be “inbound,” “outbound_ma” (manual outbound), “out_pre” (preview outbound), “conference,” or “transfer.”
- $idskill_voc_rt: Stores the identification of the agent group (skill).
- $idcampaing_voc_rt: If the interaction is an outbound call in predictive dialing, this variable stores the identification of the associated campaign.
- $idcustomer_voc_rt: Contains the identification of the customer with whom the interaction is being conducted.
- $MrWizardRes: This variable is used to store the response generated by Mr. Wizard. This response can be provided to the customer or used for future processing.
- In the following example, a “Case” component has been used to build the flow according to the type of communication channel detected in the “$channel_MrWizard” variable.
- If the channel is of type “Chat,” it will proceed to the “Artificial Intelligence Intentions” component. In its configuration, several categories have been trained to guide the flow along different paths based on what is stored in the “$last_phrase_voc_rt” variable, which holds the last sentence or phrase spoken by the customer during the conversation.
- Subsequently, using multiple “wolkvox NLP General” components, training has been carried out in each one with specific information. The purpose is for Mr. Wizard to respond to the question posed by the agent based on the topic related to what the customer indicated, as detected in the “$last_phrase_voc_rt” variable.
The same applies in the case where it is detected that the communication channel is of “Voice” type. The “wolkvox NLP General” component has been utilized to train Mr. Wizard on a specific topic, allowing it to respond to the agent with all questions related to the topic and the information provided within this component.