LocalAIAgentWithRAG

🍕 Pizza Restaurant Review AI Agent

A local AI-powered question-answering agent that uses Retrieval-Augmented Generation (RAG) to answer questions about a pizza restaurant based on real customer reviews. Runs entirely on your machine using Ollama — no API keys or cloud services required.

---

How It Works

1. Embeddings — Customer reviews are embedded using mxbai-embed-large via Ollama and stored in a local ChromaDB vector store.
2. Retrieval — When you ask a question, the top 5 most relevant reviews are retrieved using semantic similarity search.
3. Generation — The retrieved reviews and your question are passed to llama3.2 (also via Ollama), which synthesizes a grounded answer.

Question → Retrieve relevant reviews → LLM generates answer from reviews

---

Project Structure

├── knowledge_store/
│   └── realistic_restaurant_reviews.csv   # Review dataset
├── chrome_langchain_db/                   # Auto-generated ChromaDB vector store
├── vector.py                              # Embedding + vector store setup
├── main.py                                # Chat loop and LLM chain
├── requirements.txt
└── README.md

---

Prerequisites

• Ollama installed and running locally
• Python 3.9+

Pull the required models before running:

ollama pull llama3.2
ollama pull mxbai-embed-large

---

Setup

1. Clone the repository

git clone https://github.com/harnish7576/LocalAIAgentWithRAG.git
cd LocalAIAgentWithRAG

2. Create and activate a virtual environment

python -m venv venv

macOS / Linux:

source venv/bin/activate

Windows:

venv\Scripts\activate

3. Install dependencies

pip install -r requirements.txt

4. Run the agent

python main.py

The first run will embed all reviews and persist them to ./chrome_langchain_db. Subsequent runs skip this step and load directly from the vector store.

---

Demo

---

Requirements

langchain
langchain-core
langchain-ollama
langchain-chroma
chromadb
posthog
numpy
pandas

Install all at once:

pip install -r requirements.txt

---

Data Format

Reviews are loaded from a CSV file with the following schema:

Column	Type	Description
Title	string	Short title of the review
Date	string	Date of the review
Rating	integer	Star rating (1–5)
Review	string	Full review text

---

Tech Stack

Component	Tool
LLM	llama3.2 via Ollama
Embeddings	mxbai-embed-large via Ollama
Vector Store	ChromaDB
Orchestration	LangChain
Data	pandas

---

Notes

• The vector store is only populated on the first run. To re-embed (e.g. after updating the CSV), delete the chrome_langchain_db/ directory and rerun.
• All inference is local — no data leaves your machine.
• To swap the LLM or embedding model, update the model names in main.py and vector.py respectively, and pull the new models via Ollama.