Stateful AI Chatbot App Built with LangGraph, Llama 3.3

Jul 31 (edited) • 💬 General

Hey everyone,

I recorded a short video demo showing a stateful chatbot app I built using LangGraph, Streamlit, and ChatGroq.

I wanted to move beyond linear message arrays and build a structure that handles state checkpointers and thread isolation properly.

How the setup works:

Graph State: Custom state schema defined with message reducer logic.
Persistence: Uses SqliteSaver checkpointers to save and retrieve state snapshots across sessions.
Streaming: Streamlit consumes the workflow message stream directly to render responses token-by-token.
Interface: Custom layout built in Streamlit for thread switching and active conversation previews.

Check out the video and let me know if you have any questions about wiring LangGraph streams into a Streamlit UI.

2 comments

Stateful AI Chatbot App Built with LangGraph, Llama 3.3