Practical overview
When it fits
Use it when you need a learning-oriented starting point or an extensible prototype for LangGraph-based RAG with conversation context and human input during query clarification.
Who it is for
- People learning agentic RAG concepts
- Developers prototyping document-based conversational assistants
- Experimenters exploring memory, retrieval orchestration and human clarification
Documented uses
- Learn how an agentic RAG workflow is structured
- Build a conversational question-answering prototype over documents
- Clarify ambiguous or multi-part user queries before retrieval
- Experiment with modular LLM, embedding, document-processing and workflow components
Requirements & limits
The repository is public and MIT licensed. It is presented for desktop or self-hosted use, with an interactive notebook path and a runnable application described as Ollama-first.
- The project offers both a step-by-step notebook path and a modular building path.
- Its retrieval design uses small child chunks for search and larger parent chunks for answer context.
- Conversation memory uses a rolling summary and recent history rather than indefinitely expanding context.
- Complex questions can be decomposed into parallel sub-queries through a multi-agent map-reduce workflow.
- Ollama is the documented default for the runnable application, while LangChain provider examples include Anthropic, OpenAI and Google.
Check current requirements in the official documentation. Source-based guidance does not mean OpenViST has run or independently tested this project.
Open the interactive Agentic RAG learning notebook
Begin exploring the documented step-by-step Agentic RAG tutorial.
Before you start
- Access to the public GitHub repository.
- A notebook environment; the repository provides an Open in Colab link to the notebook.
Steps
Open the repository’s linked agentic_rag.ipynb notebook.
Follow the notebook learning path to explore the core Agentic RAG concepts.
Use the notebook as the starting point before moving to the modular project path.
Expected outcome
A guided starting point for understanding the project’s LangGraph-based agentic RAG workflow.
Guide limitations
- The supplied material does not document complete execution steps or a verified run.
- This guide covers the learning path, not deployment of the modular application.
Sources & review
Source record details
- Source availability
- Active
- Sources checked
- 1 October 2026
- Original website title
- GiovanniPasq/agentic-rag-for-dummies · Official homepage title · retained in its source language