In this chapter, we looked at building chatbots or neural conversational agents using neural networks and deep learning. We first saw what makes a chatbot and the main forms in use today: goal-oriented and conversational bots. Then we looked at how to build a basic machine translation conversational chatbot that used sequence-to-sequence learning. 

After getting a background in sequence learning, we looked at the open source tool DeepPavlov. DeepPavlov is a powerful chat platform built on top of Keras and designed for many forms of neural agent conversation and tasks. This made it ideal for us to use the chatbot server as a base. Then we installed RabbitMQ, a microservices message hub platform that will allow our bot and all manner of other services to talk together later on.

Finally, we installed Unity and then quickly installed the AMQP plugin asset and connected to our chatbot server.

This completes our introductory section to deep learning, and, in the next section, we begin to get more into game AI by diving into deep reinforcement learning.