Hi everyone!
I’ve been learning LangGraph for a while now and I think it’s a fantastic and approachable framework (I actually went through the entire course three times 
). To practice, I built a revisited version of the research assistant from Chapter 4—a more structured and modular take on the original.
One thing I realized is that having a good sense of system architecture is crucial when building agents, especially for designing state flow—how state is shared, isolated, or transitioned between agents. Honestly, this was one of the hardest parts for me.
I initially promised myself not to use any AI help, just to really understand things in depth… but I caved! I’m only human (and yes, I know: using AI the right way is key to not getting lost—it’s hard sometimes).
I’d love to hear from you:
-
How do you design your agents and their states?
-
For multi-agent systems, do you follow any patterns for state management or agent orchestration?
-
Any feedback on my approach—what could I have done better, or what did I get wrong?
I’d really appreciate any advice or critique. Thanks for building such a great community and tool!