Observe AI

Agentic RAG System using LangChain, LangGraph, and Gemini API

Demo Video

ObserveAI_DemoVideo.mp4

Requirements

• Python 3.11
• Google Gemini API Key

Setup Instructions

1. Clone the Repository

git clone https://github.com/PriyamPritamPanda/Observe.git
cd Observe

Do not forget to be on the right branch

2. Create a Python Virtual Environment

python3.11 -m venv venv
source venv/bin/activate   # On Windows: venv\Scripts\activate

3. Install Dependencies

pip install -r observe_ai/requirements.txt

4. Create a .env File in the Root Folder

Create a file named .env in the root directory and add your Gemini API key:

GEMINI_API_KEY=your_google_gemini_api_key_here

Folder Structure

Observe/
├── .github/
├── archived/
├── observe_ai/
│   ├── requirements.txt
│   ├── src/
│   │   ├── main.py
│   │   └── ...
├── venv/
├── .env
├── .gitattributes
├── .gitignore
├── README.md

Running the Application

Change directory to observe_ai and run the main module:

cd observe_ai
python -m src.main [ARGS]

Main Arguments

• --init-vectors : Initialize vector store from transcripts
• --interactive : Run in interactive mode (CLI)
• --query "your question" : Process a single query

Examples

Initialize vector store:

python -m src.main --init-vectors

Run in interactive mode:

python -m src.main --interactive

Process a single query:

python -m src.main --query "What was the outcome of transaction 123?"

Notes

• Ensure your .env file contains a valid GEMINI_API_KEY before running.
• Python 3.11 is required for compatibility.
• All commands should be run from the observe_ai directory.
• src/memory and src/causal are redundant for now. Needs work.