⚖️ DebateBot — Multi-Agent AI Reasoning System

DebateBot is an advanced multi-agent AI reasoning platform that takes any complex topic or decision, runs it through opposing advocate agents, and delivers a rigorous, objective final judgment. Built for the Microsoft Agents League Hackathon 2026 — Reasoning Agents Track.

🔗 Live Demo: debatebot-adhipatya.streamlit.app

---

🤖 Multi-Agent Architecture & Flow

Instead of a single LLM prompt, DebateBot simulates a structured debate panel where agents are configured with specialized roles and optimal inference parameters (temperatures) to maximize reasoning depth:

                  ┌─────────────────────────────┐
                  │      User Input Topic       │
                  └──────────────┬──────────────┘
                                 │
                 ┌───────────────┴───────────────┐
                 ▼                               ▼
      ┌───────────────────────┐       ┌───────────────────────┐
      │   Agent 1: FOR        │       │   Agent 2: AGAINST    │
      │   (Temp: 0.85)        │       │   (Temp: 0.85)        │
      │   Llama 3.3 70B       │       │   Llama 3.3 70B       │
      └───────────┬───────────┘       └───────────┬───────────┘
                 │                               │
                 └───────────────┬───────────────┘
                                 ▼
                      ┌───────────────────────┐
                      │    Agent 3: JUDGE     │
                      │    (Temp: 0.15)       │
                      │    Llama 3.3 70B      │
                      └───────────┬───────────┘
                                 ▼
                     ┌───────────────────────┐
                     │ Impartial Verdict     │
                     │ + Confidence Score    │
                     └───────────────────────┘

1. Agent FOR (Creative & Persuasive): Instantiated with temperature=0.85 to generate 5 sharp, bold, evidence-backed arguments supporting the topic.
2. Agent AGAINST (Creative & Critical): Instantiated with temperature=0.85 to formulate 5 distinct counter-arguments, citing statistics and real-world impact.
3. Agent JUDGE (Analytical & Impartial): Instantiated with a cool temperature=0.15 to ensure consistent logic, identify blind spots missed by both sides, assign a confidence score, and deliver an objective verdict.

---

⚡ Tech Stack (100% Free)

Layer	Tool / Model	Cost	Details
LLM Inference	Llama 3.3 70B / Llama 3.1 8B via Groq	$0	Lightning-fast inference (no token latency)
Agent Framework	LangChain	$0	Handles chains, templates, and multi-agent routing
Frontend UI	Streamlit	$0	Elegant, high-performance interactive interface
Deployment	Streamlit Cloud	$0	Live-hosted web app at debatebot-adhipatya.streamlit.app

---

🚀 Local Quickstart

Prerequisites

• Python 3.10 or higher
• A free Groq API key (obtained from console.groq.com with no credit card required)

Setup Instructions

1. Clone the Repository

   git clone https://github.com/adhipatya3552/DebateBot.git
   cd DebateBot

2. Create & Activate Virtual Environment

   • Windows:

     python -m venv venv
     venv\Scripts\activate

   • macOS/Linux:

     python3 -m venv venv
     source venv/bin/activate

3. Install Dependencies

   pip install -r requirements.txt

4. Launch the Application

   streamlit run app.py

   Open http://localhost:8501 in your browser.

---

🏆 Hackathon Submission Checklist

• Multi-agent reasoning pipeline showing reasoning and multi-step thinking (Reasoning track requirement)
• Zero credit-card / paid subscriptions required (100% Free Stack)
• Clean, responsive, glassmorphic layout styling with gradient branding
• Downloadable debate results as Markdown records for B2B/academic auditing
• Supporting README explaining the project structure and setup