It's time! The Chatbot can officially be tested. For this we have developed the Trengo Playground, in which the Chatbot can be optimally tested for present knowledge and answers. This way the added articles and answers can be checked by all kinds of questions.
How does the Playground work?
If a question is asked in Playground, Trengo gives an answer if it's matched to it. If a match is found, then it will be stated exactly how many percent it's a match.
In the case above, the Chatbot is 100% convinced that the question "Good morning!" fits with the answer "You also have a good morning".
An uncertain match
If the matching percentage is shown in red, the Chatbot will only give this answer as a suggestion in Playground. The percentage comes under the previously set answer confidence percentage, which is set under "Settings". In the live environment (in which the customer will ask all his/her questions) the Chatbot will say that he does not understand the answer.
If no related answer is found, the Chatbot in Playground will answer "I am a chatbot and I still have to learn a lot. Could you ask this question in a different way?" (100%).
Complex questions in Playground
With complex questions, each sentence of the question will be matched to an answer and the question will be divided into answers with different matching percentages. In the example below we explain how Playground deals with this situation.
In the situation below, the Chatbot will provide an answer to this complex question in real life:
Because complex questions are divided into individual (small) questions, there's a big chance that the main question will be answered. In addition, the Chatbot also indicates what he hasn't understood (yet).
In this case, the person who asks the question understands that it's not necessary to ask "my name is Derk" and "I like to hear it" again. This is not the intention of the message. Now, he/she knows that he/she can report a malfunction via email and is immediately helped.
Have you tested the Chatbot sufficiently and are ready for the Chatbot to cooperate in your team ?