Finetune model with conversations

I have read about the advantages of fine-tune models for specific cases. However, all the examples mentioned in the different courses where about summarization and other use cases. I am looking to know if it is possible to do it with, for example, old Slack conversations. The idea with this fill be to let the model know how to answer and how to follow the same structure for the conversation. Example:
#Person 1: I want to know if it is raining today
#Person 2: In order to know that I need to know where are you located
#person 1: I am in New York
#Person 2: Then there is a probability to rain today

In this scenario, the goal after finetune must be to let the model act as the Person 2

Yeah you could do it (I havent done it myself) as long as you have the conversation you could fine tune the LLM with that dataset in the same way that was done in the course.