Your business and communication processes are the basis for the chatbot. After all the system is intended to do what you want it to do and nothing else. In order to make this happen, we proceed in the following steps:
- Translation of communication processes into chat flows
- Clear definition of questions asked by the chatbot and answer options
- Use of import, export, and simulation tools
In order to not only communicate in fixed, pre-defined workflows but also be able to identify free text customer requests, we add intelligence to the chatbot based on NLP and NLU.
- Based on Natural Language Processing (NLP) a customer's message is disassembled into its linguistic components
- Thanks to Natural Language Understanding (NLU) the system identifies the customer's intent and starts the corresponding process
- Moreover, not only the intent but also all other relevant information is extracted from the customer's message. If the extracted information already answers questions the chatbot was going to ask, those questions are skipped.
If the chatbot does not recognize the customer's intent, the system uses the predefined workflow as a fallback and leads the customer through the predefined set of questions. After the workflow is completed, the customer's initial message and the answers, the customer gave in the course of the workflow, are connected to create new so called "search paths" to the system's AI, enabling the system to identify the newly learned customer intent in the future.