README.md

metadata

title: RAGent Chatbot
emoji: 🤖
colorFrom: blue
colorTo: green
sdk: gradio
sdk_version: 5.38.0
app_file: app.py
pinned: false
license: mit
short_description: A Smart AI chatbot powered by RAG and AGENT

💬 Ragent Chatbot Preview

🤖 Ragent Chatbot

Ragent Chatbot is an intelligent retrieval-augmented agent assistant powered by LLMs. It combines the power of RAG (Retrieval-Augmented Generation) with agent-based tool reasoning, allowing it to dynamically respond using retrieved knowledge or external tools depending on the query.

🧠 What It Can Do

• Answer user questions by retrieving information from a custom document store
• You can upload any document and then ask the chatbot to retrieve answer of your question.
• Automatically decide when to use tools (like search, calculator, etc.)
• Combine multiple steps of reasoning using the ReAct agent pattern

🛠 Features

• 🔎 Hybrid Search: Combines vector similarity and BM25 keyword matching for relevant document retrieval
• 🤖 ReAct Agent: Uses tool-based reasoning when knowledge retrieval is insufficient
• 💬 Gradio Chat UI: Simple and responsive chat interface
• 🧱 Modular Tools: Easily extendable with tools like web search, calculator, and custom APIs

📦 Stack

• Frontend: Gradio
• Agent Framework: LangChain (ReAct agent)
• Vector DB: Qdrant
• LLM: Gemini
• Embedding Model: BAAI/bge-large-en-v1.5

🔍 Example Queries

Try asking questions like:

1. "What is LangChain and how is it different from LlamaIndex?"
2. "Who is the CEO of OpenAI and when was the company founded?"
3. "What is 245 * 92?"

💡 The chatbot decides whether to use RAG or call tools like calculator or web search automatically!

🔗 GitHub Repository

You can explore the full source code, Docker setup, and implementation details on GitHub:

👉 RAGent Chatbot

🚀 Try It Live

• Click url to launch the app in Hugging Face: RAGent Chatbot