Hi All,
I am building an chatbot, that has intake, symptom and appointment agent
a) Intake agent is responsible to collect answer from user for 14 questions, save it into the database and also at any point of time user wants to update the intake details it will do
eg: Height, Weight, Age, blood group
b) Symptom agent is responsible to collect user sympto
c) Appointment agent is responsible to collect appointment details
These three agent is part of single chatbot and user must complete intake before accessing symptom and appointment agent
User can ask any question in the conversation
eg: Hello, I want to update my weight,
So, I am not sure how do i build this in langgraph suggestions would be helpful
hi @Syed_Mujeeb
nice project!
I would model this as a single LangGraph with:
- a router node that detects intent
- three domain subgraphs/nodes (intake, symptom, appointment)
- gating via conditional edges that route users back to intake until it’s complete.
Use a checkpointer for persistence, and interrupts when you need to pause for human confirmation or missing info. That gives you multi‑turn, resumable flows and clean boundaries between responsibilities.
If you want a supervisor approach instead of a simple router, wrap each domain agent as a subgraph and drive them with a top-level supervisor node. The same gating logic still applies - just route to intake until intake_completed is true - then allow symptom/appointment.
Hi @Syed_Mujeeb !
Fur further inspiration, here are our docs on multi-agent architectures. You can also consider using our new deepagents library!