LangGraph: Best way to collect sequential user input for appointment booking?

[areebaayub29]

Question:
I am creating a multi turn customer service bot which have different responsibilities,
One of those responsibility is book an apppointment,
We have created a node where we have defined a tool to schedule appointment, that tools expect (date,time location) parameters.
But we want to create a flow where bot will first ask about date then on the basis of that date will fetch available slots and ask user to select it and then location. (al of them sequentially)
But we are reluctant on how to do this flow in a best possible way, where there will be less llm call and proper structured flow,
One of the solution came to our mind is , llm willl call the tool (appointmentnode) and return all the missing information then we will collect all those missing data from user manually (through interrupt on other node , maintain a state and send back those information to llm again (appointment node) to book and appointment,
Can anyone help me in creating proper flow,

Thankyou.

[hinthornw]

Depending on how complicated the form is, you could do somethign like this: https://langchain-ai.github.io/langgraph/tutorials/chatbots/information-gather-prompting/

You could use an interrutp in the final appointment scheduling node as a "final confirmation", but before that I'd probably present to the model the expected fields to collect for the flow all in the system prompt and let hte model prompt the user for it

[areebaayub29]

my proposed solution:

[olivia2801]

You can look unto Agentra.io for the same as it also has a feature of AI appointment scheduling.