PRISM | Multi-Agent AI Debate Platform

An AI debate platform where multiple LLM agents research, deliberate, challenge each other, and produce a final synthesized answer.

PRISM is a sophisticated multi-agent AI platform that orchestrates six AI agents through multiple rounds of debate and deliberation. The system combines web research, semantic memory, real-time communication, and AI synthesis to generate higher-quality answers than a single model. Built as a full-stack application with authentication, persistent conversations, and live streaming responses.

Tech Stack

Next.jsReact 19Node.jsExpress.jsMongoDBMongooseSocket.IOGeminiGoogle OAuthJWTTavily

Features

6-agent AI debate orchestration system
2-round deliberation and challenge mechanics
Real-time streaming responses with Socket.IO
Semantic memory using Gemini embeddings
Web research integration with Tavily
Google OAuth 2.0 authentication
JWT authentication with HttpOnly cookies
Persistent conversation history

Challenges

Coordinating multiple AI agents in a structured debate workflow.
Managing real-time communication between agents and users.
Implementing semantic memory and contextual retrieval.
Ensuring efficient orchestration while maintaining response quality.

Learnings

Advanced AI agent architecture design.
Semantic search and embeddings implementation.
Real-time event-driven systems with Socket.IO.
Building production-ready authentication systems.

Feedback

For feedback or suggestions, contact me at: shahzaibalijamro@gmail.com

Links

Live Github