Argu AI - Debate Platform

Argu AI is an AI-powered debate platform built for the ModusHack Hackathon. It allows users to choose debate topics, assign AI personas, and watch the models engage in real-time debates.

This project leverages Modus for orchestrating AI models and Go as the programming language, making it an interactive and real-time AI debate experience.

---

🚀 Project Overview

Argu AI is designed to facilitate engaging and interactive AI debates. Users can select two AI models from multiple options (e.g., Gemini, GPT-4, Meta LLaMA) and assign them different personas. The debate occurs in a back-and-forth format, where the models argue different viewpoints on a given topic. After 8 rounds of debate, a third AI model acts as the judge to decide the winner based on the debate performance.

---

💻 Technical Details

Modus for Model Orchestration

Modus is used to seamlessly orchestrate the communication between multiple AI models. The output from one model is fed as input into the next, creating a fluid and continuous debate process.

Real-Time Data Storage

Debate messages arestreamed and stored in PostgreSQL (via Azure Cosmos DB) for real-time updates. This ensures that frontend can track the ongoing debate without significant delays.

---

📸 Debate Flow

Here’s how the debate works internally:

1. User Inputs Topic: The user selects a topic for the debate.
2. Model Selection: Two AI models are selected, each with a unique persona.
3. Debate Flow: The models debate in real-time, alternating their arguments.
4. AI Judge Evaluation: After 8 rounds, a third AI model judges the debate.
5. Winner Announcement: The winning model is decided based on the judge's evaluation.

---

⚙️ Setup AI Models

Note: For the hackathon, the available models are limited due to API rate limits. Please ensure to use the Gemini model as it has higher rate limits..

⚡ Getting Started

To run this project locally, follow these steps:

1. Prerequisites

Before running the project, ensure you have the following installed and configured:

• Go and TinyGo
• Install the Modus CLI globally with npm:

npm install -g @hypermode/modus-cli

Cloning the Repository

git clone https://github.com/satti-hari-krishna-reddy/arguai

cd arguai

Next, replace the environment variables in .env.dev.local with your credentials:

MODUS_OPENAI_API_KEY=
MODUS_GOOGLE_API_KEY=
MODUS_CONVDB_PG_USER=
MODUS_CONVDB_PG_PASSWORD=

Run the App Locally

To run the app use the following command:

 modus dev

This will start your app at http://localhost:8686/graphql.