Generative AI in task based chat bots

I am a software developer with little experience in practical AI. My problem statement is I need to develop a POC of a chatbot that helps users complete a task. In this case choose from a set of cars. There are a lot of different cars offered by a company with a lot of add ons. It becomes difficult for the customer to consume all the details and hence comes the agent in picture. I want to test the hypothesis that chatbots that feel like human would help narrow the choices for the customer faster hence decreasing the purchase period.

From what I have studied so far I the problem statement can be broken down into two parts a supervised part that handles intent recognition, NER and an unsupervised part that handles the text that is being sent as an output.

Is this assessment correct?
Also can somebody suggest me reading resources for this.