GamerMajlis Chatbot

Welcome to the GamerMajlis Chatbot repository! This project is a mini-RAG-based chatbot system designed for the gaming platform GamerMajlis, aimed at enhancing user experience and answering platform-related questions efficiently.

---

🚩 Platform Overview

GamerMajlis is a gaming platform that allows users to connect with other gamers, buy gaming gear, participate in events and tournaments, communicate seamlessly and access all gaming-related services in one place

The chatbot was designed to simplify user interactions and provide immediate answers to both static and dynamic questions.

---

🚩 Repository Structure

Here’s a breakdown of the files and their purpose:

File / Folder	Description
chatbot.py	Main chatbot script that initializes without the Gradio UI and connects the system.
chatbot_gradio.py	Chatbot script that with the Gradio UI.
db_connection.py	Database connection logic for dynamic queries.
queries.py	Maps intents to SQL queries for dynamic platform questions.
static_intents.py	Dictionary of static intents and their responses.
system_prompt.txt	LLM prompt that sets its role and rules, including “no fake data.”

🚩 How the Chatbot Works

The system relies on a built-in LLM called meta-llama/llama-4-scout-17b-16e-instruct, which is responsible for understanding queries and generating responses.

Question Types

The chatbot can handle three main types of user questions:

1. General Questions

   • Not related to the platform.
   • Directly answered by the LLM.

2. Dynamic Platform Questions

   • Questions about events, tournaments, or products (dynamic data that changes).
   • The chatbot first identifies an intent from the question.
   • Each intent is mapped to a specific database query.
   • The result is formatted and sent back to the LLM for a user-friendly response.

3. Static Platform Questions

   • Questions that don’t change, e.g., “How do I change my password?”
   • Handled via a predefined dictionary called static_intents.py.
   • LLM provides natural responses based on these static intents.

---

🚩 Key Features

• Intent Extraction – Each question generates an intent key to map queries securely.
• Dynamic Queries – Queries only run when needed, minimizing unnecessary database access.
• Static Intents – Common, unchanging questions are answered efficiently without querying the database.
• Safety Prompting – The LLM never generates fake data. If information is missing, it explicitly says so.
• Customizable UI – Built using Gradio, fully styled with CSS for a smooth user experience.

---

Note: The .env file contains sensitive keys and is ignored in Git.

---

🚩 How to Run Locally

Follow these steps to get the project running on your machine:

# Clone the repository
git clone https://github.com/GamerMajlis-platform/AI.git
cd AI

Configure Environment

1. Create a .env file with the following structure:

GROQ_API_KEY=<your_groq_api_key>
DATABASE_NAME=<your_db_name>
DATABASE_USER=<your_db_user>
DATABASE_PASSWORD=<your_db_password>

2. Ensure your database matches the queries in queries.py, or customize it.

3. Create a Gradio API key if required.

4. Run the chatbot:

python chatbot.py

Your Gradio UI will launch locally, ready for testing.

---

🚩 Customization Tips

• Dynamic Tables / Queries: Adjust queries.py to match your database structure.
• Static Intents: Modify static_intents.py for additional FAQs.
• LLM Prompt: Update system_prompt.txt to guide the chatbot behavior.
• UI Customization: Update CSS in chatbot.py to adjust colors, sizes, or layout.

---

Database Query Flow

• Dynamic question → intent from queries → query → LLM formatting → response
• Static question → intent from ststic → query → LLM formatting → response
• Other question → LLM response