Building Chatbots

In the previous chapters, we discussed how to answer questions on long documents using summarization techniques. However, answering a single question may not be enough for your users. Enterprise question-answering is one of the most popular use cases these days, and users expect multiple turns in their conversation. In this chapter, we will discuss how we can leverage the power of LangChain to add memory to our LLM system, turning it into a powerful chatbot. Before we dive into the nuts and bolts, let’s discuss the experience our users would like to have with a chatbot.

ChatGPT and other GenAI-powered chatbots have shaped user experience expectations toward chatbots. Most chatbots have a specific purpose and are designed to help users reach a specific goal such as booking a flight, making a restaurant reservation, or answering difficult questions. Users also want their chatbot to be conversational, with a friendly and approachable demeanor. And finally,...