README.md

metadata

title: ARGObot
emoji: 🤖
colorFrom: indigo
colorTo: gray
sdk: streamlit
sdk_version: 1.32.2
app_file: app.py
pinned: false

ARGObot: UWF's Custom AI Advisor

ARGObot is a modular question-answering chatbot that uses either OpenAI (GPT-4) or Google's Gemini with Retrieval-Augmented Generation (RAG). Designed with Langchain, LangGraph, and Streamlit, it allows students to query university policies directly from documents like the UWF Student Handbook.

Features

• Easily switch between OpenAI and Gemini models.
• Retrieval-Augmented Generation (RAG) using PDF knowledge base.
• Tool integration for fallback search (Google Search).
• Conversational memory with LangGraph.
• Custom UI with UWF branding and Streamlit.

Local Development

# Install dependencies
pip install -r requirements.txt

# Set environment
export MODEL_PROVIDER=openai  # or gemini
export OPENAI_API_KEY=your-key
export GOOGLE_API_KEY=your-key

# Run Streamlit
streamlit run app.py

Project Structure

.
├── app.py                  # Unified entrypoint
├── requirements.txt
└── src/
    ├── agents/             # Prompt templates and tool config
    ├── chains/             # OpenAI & Gemini RAG logic
    ├── config/             # Environment setup
    ├── interface/          # Streamlit UI
    └── state.py            # LangGraph definition

Hugging Face Secrets (Required)

Add the following secrets in the Hugging Face UI:

• OPENAI_API_KEY
• GOOGLE_API_KEY
• MODEL_PROVIDER (default: openai)