Radiance SmartChat, brand safety, and chatbot responses

As powerful and exciting the new AI chatbots like ChatGPT are, merchants don’t want to put a ChatGPT chatbot directly on their website. That’s because they want to control the behavior of their chatbot. In addition to designing the other aspects of the chatbot experience, they want their chatbot to use their knowledge base, and only their knowledge base, to answer user questions. That’s one of the many value-adds of Radiance SmartChat.

Radiance SmartChat uses OpenAI’s GPT API to generate responses to user’s input, but it guides GPT to only use the relevant pieces of information from the merchant’s Knowledge Base which includes the product catalog, FAQs, order history etc. This is to ensure that the responses are correct, appropriate, and also brand safe.

While the modern AI is amazing in generating creative, delightful, human-like answers to user’s questions most of the time, it is also known to make mistakes. Since most of our answers are generated in real-time by GPT, it is impossible to guarantee the correctness or brand-safety of all the answers. We include  disclaimer at the bottom of our widget to alert the user about potential inaccuracies of the answers.