Unit Tests. Onboarding Strategy

The outcome of the development stage is a software product that is ready for QA. In this chapter, we will focus on specific activities related to AI chatbot development. We would assume that the reader is familiar with industry standards in software development, and thus does not need to focus on that.

In this video, the speaker considers how to define a client’s onboarding strategy and unit tests. This provides some inspiration regarding possible sources in the video.

The bot architecture could be rather complex and it will have an integration with the many other systems that are generally connected via Rest APIs

One of the issues with such architecture is that the API could be changed and the user (bot or other system components) could not be notified about said change, and that could lead to issues that are hard to predict. That is specifically relevant when the component is not mature and its development is still in process.  

So our recommendation would be to cover the API with unit tests (in case you do not own it), so any issues related to changes in API could be quickly identified and addressed.

The majority of the bots fail to recognize intents and thus respond appropriately to a user, so the creation of a correct test case to validate bot behavior is vital for project success. While developing standard positive and negative test cases, we invite you to consider the cases for dialog testing we presented in video.  


See the next lesson Test Cases