Hi everyone!
I’m building a system which provides 5 types of recommendations. The user experience will be of a wizard, when I’ll be asking 20-30 questions one be one, when each question might be skipped, asked or updated based on the previous answer.
The questions are grouped based on the type of recommendations, but there might be an overlap of questions for each recommendation. For example. I’ll ask questions 6-10 and then 25 in order to generate recommendation A, and ask 8-15 and 25 to generate recommendation B.
I’m thinking about building a data collection agent which is based on the Assistant API and will ask each question, and based on each instruction continue to the next, skip it, or use the answer of previous question as the answer to this question as well.
It’s the first time I’m building with the assistant api and the terms and components are a bit unclear.
From what I understand, I need to create an assistant and then a thread.
- Will each “message” in the api be a question?
- Where do I place the instructions for the following question?
- For example, when I’m asking question 21, how can I present the user with the answer to question 3 and ask him/her to confirm it or update it?
Any guidance would be greatly appreciated!