Agentic RAG: Build a multi-agent QA chatbot for your documents with RAG, LangGraph and Docling

Channel:

IBM Developer

Subscribers:

74,100

Published on June 4, 2025 7:55:21 PM ● Video Link: https://www.youtube.com/watch?v=12x0vDfAYA4

Duration: 0:00

143 views

3

In this immersive workshop, you'll build DocChat, a multi-agent RAG platform that retrieves, synthesizes, and verifies information from your documents with precision. Unlike traditional chatbots, DocChat combines advanced hybrid retrieval (BM25 + vector search) and multi-agent workflows to cross-check answers. No more hallucinated responses—DocChat delivers fact-based solutions or flags gaps when data is missing.

Using LangGraph, Docling, ChromaDB, and Gradio, you’ll learn to deploy reliable AI that excels at real-world document intelligence. Walk away with practical skills to build AI assistants that are transparent, accurate, and production-ready.

Check the following repo for the Model Context Protocol code from the video, updated and production ready: https://ibm.biz/haileys-repo

Like this session? Want more? Check out presenter Hailey Quach's free guided project! https://ibm.biz/agentic-rag-guided-project



Connect with developers and other practitioners in the IBM TechXchange Community: https://ibm.biz/techxchange-community-llmdd

Choose a specific community topic group:

watsonx.ai: https://ibm.biz/watsonx-ai-llmdd
watsonx Code Assistant: https://ibm.biz/watsonx-code-assistant-llmdd
IBM Granite LLMs: https://ibm.biz/watsonx-ai-llmdd
watsonx.governance: https://ibm.biz/watsonx-governance-llmdd


For more technical tutorials, articles, and learning resources, check out IBM Developer: https://ibm.biz/ibm-developer-yt

Subscribe to see more developer content: https://ibm.biz/ibm-developer-yt-subscribe

Follow IBM Developer on LinkedIn: https://ibm.biz/ibm-developer-linkedin-yt


#programming
#IBMTechXchange
#IBMDeveloper
#Developer
#Coding