Build a Dialogue Chatbot using Azure OpenAI and Langchain

Ferry Djaja
10 min readAug 30, 2023

In this blog post, I would like to share what I did to come out with a conversational ChatBot during my hackathon where I created a conversational ChatBot with Azure OpenAI and LangChain.

Initiating a conversation with the bot is triggered by the user. The conversation revolves around the topic of requesting leave from work, and the bot’s objective is to gather the subsequent information from the user:

  • When is the user planning to go on leave?
  • Which country is the user intending to visit?
  • Is a visa letter required for the user’s travel?
  • Does the user require assistance in configuring an out of office message?
  • Is the user seeking guidance on delegating their approval authority to someone else?

After the bot has received the user’s responses, it will proceed with the next steps to configure the out of office message user’s calendar and so on and so forth.

Here are some conversation screenshots. In the given snippet of dialogue below, the bot’s aim is to gather essential information before moving forward to the next phases. The bot will verify that it has collected all the essential particulars and organize them within a JSON structure.

However, consider a scenario where the user does not mention the country they intend to visit during the conversation; in such cases, the bot will prompt this question.

Yet, if the user has furnished all the essential information, the bot will simply respond with instructions to proceed.

I trust this provides you with a grasp of the concept we are aiming to build.

Basic Flow