DeepR - AI Decision Frameworks Demo

A demo-ready React SPA showcasing three AI decision-making frameworks:

1. LLM Council - Multiple AI models deliberate and a chairman synthesizes
2. DxO Decision Orchestrator - Role-based sequential decision-making
3. Adversarial Debate - Advocate-Challenger-Arbiter pattern for refined reasoning

Features

LLM Council Framework

• Multiple AI council members provide independent answers
• Each member shows confidence scores and detailed reasoning
• Chairman agent synthesizes all inputs into final recommendation
• Visual indicators for consensus and disagreements

DxO Decision Orchestrator

• Role-based sequential analysis (Lead Researcher → Critic → Expert → Analyst)
• Each role builds upon previous roles' outputs
• Final decision shows revisions and how feedback was incorporated
• Inspired by Microsoft's MAI-DxO research

Adversarial Debate Framework

• Three-role pattern: Advocate → Challenger → Arbiter
• Multi-turn debate cycles with iterative refinement
• Smart convergence detection or configurable turn limits
• Adversarial critique strengthens final reasoning

Tech Stack

• Frontend: React 18 + Vite + Tailwind CSS
• Backend: Node.js + Express
• LLM Integration: Multi-provider support (OpenAI, Anthropic, Google Gemini, OpenRouter) with mock mode for demos

Quick Start

Prerequisites

• Node.js 18+
• npm or yarn

Installation

# Clone the repository
git clone <repo-url>
cd DeepR

# Install dependencies
npm run install:all

Configuration

1. Copy the environment example file:

cp backend/.env.example backend/.env

2. Configure your settings in backend/.env:

# Use mock data (no API key needed)
USE_MOCK=true

# Or use real AI APIs
USE_MOCK=false

# OpenAI API Configuration
AI_INTEGRATIONS_OPENAI_API_KEY=your-openai-api-key-here
AI_INTEGRATIONS_OPENAI_BASE_URL=https://api.openai.com/v1

# Anthropic API Configuration (optional)
AI_INTEGRATIONS_ANTHROPIC_API_KEY=your-anthropic-api-key-here
AI_INTEGRATIONS_ANTHROPIC_BASE_URL=https://api.anthropic.com

# Google Gemini API Configuration (optional)
AI_INTEGRATIONS_GEMINI_API_KEY=your-gemini-api-key-here
AI_INTEGRATIONS_GEMINI_BASE_URL=https://generativelanguage.googleapis.com

# OpenRouter API Configuration (optional, for open-source models)
AI_INTEGRATIONS_OPENROUTER_API_KEY=your-openrouter-api-key-here
AI_INTEGRATIONS_OPENROUTER_BASE_URL=https://openrouter.ai/api/v1

Running the Application

# Start both frontend and backend
npm run dev

This starts:

• Frontend: http://localhost:5173
• Backend: http://localhost:3001

Individual Services

# Frontend only
npm run dev:frontend

# Backend only
npm run dev:backend

Project Structure

DeepR/
├── frontend/               # React frontend
│   ├── src/
│   │   ├── components/    # UI components
│   │   │   ├── CouncilView.jsx
│   │   │   ├── DxOView.jsx
│   │   │   ├── AdversarialView.jsx
│   │   │   ├── CouncilMemberCard.jsx
│   │   │   ├── ChairmanCard.jsx
│   │   │   ├── RoleCard.jsx
│   │   │   ├── FinalDecisionCard.jsx
│   │   │   └── ...
│   │   ├── App.jsx
│   │   └── main.jsx
│   └── tailwind.config.js
│
├── backend/               # Node.js backend
│   └── src/
│       ├── index.js       # Express server
│       ├── config.js      # Configuration
│       ├── llmService.js  # LLM API integration
│       └── mockData.js    # Mock responses
│
└── package.json           # Root package.json

API Endpoints

Health Check

GET /api/health

Council Framework

POST /api/council
Body: { "question": "Your research question" }

POST /api/council/stream  # Streaming responses
Body: { "question": "Your research question" }

DxO Framework

POST /api/dxo
Body: { "question": "Your decision problem" }

POST /api/dxo/stream  # Streaming responses
Body: { "question": "Your decision problem" }

Adversarial Debate Framework

POST /api/adversarial/stream  # Streaming responses
Body: { 
  "question": "Your research question",
  "roles": [
    { "id": "advocate", "model": "gpt-5.3", ... },
    { "id": "challenger", "model": "gpt-5.3", ... },
    { "id": "arbiter", "model": "gpt-5.3", ... }
  ],
  "turnLimit": "smart" // or "1", "2", "3"
}

Configuration

Backend Environment Variables

Variable	Description	Default
PORT	Server port	3001
USE_MOCK	Use mock data instead of real API	true
AI_INTEGRATIONS_OPENAI_API_KEY	OpenAI API key	-
AI_INTEGRATIONS_OPENAI_BASE_URL	OpenAI API base URL	-
AI_INTEGRATIONS_ANTHROPIC_API_KEY	Anthropic API key	-
AI_INTEGRATIONS_ANTHROPIC_BASE_URL	Anthropic API base URL	-
AI_INTEGRATIONS_GEMINI_API_KEY	Google Gemini API key	-
AI_INTEGRATIONS_GEMINI_BASE_URL	Gemini API base URL	-
AI_INTEGRATIONS_OPENROUTER_API_KEY	OpenRouter API key (for open-source models)	-
AI_INTEGRATIONS_OPENROUTER_BASE_URL	OpenRouter API base URL	-

Switching Between Mock and Live Mode

In backend/.env:

# For demos without API costs
USE_MOCK=true

# For real LLM responses (requires API keys)
USE_MOCK=false
AI_INTEGRATIONS_OPENAI_API_KEY=sk-...
AI_INTEGRATIONS_ANTHROPIC_API_KEY=sk-ant-...
AI_INTEGRATIONS_GEMINI_API_KEY=...
AI_INTEGRATIONS_OPENROUTER_API_KEY=sk-or-...

Development

Adding New Council Members

Edit backend/src/config.js:

councilMembers: [
  { id: 'new-model', name: 'New Model', model: 'gpt-5.3', provider: 'Provider Name', color: '#hex' },
  // ...
]

Note: The application supports multiple AI providers including OpenAI, Anthropic, Google Gemini, and OpenRouter (for open-source models like Llama, DeepSeek, etc.).

Customizing DxO Roles

Edit the role configuration in backend/src/config.js or modify prompts in backend/src/llmService.js.

Demo Tips

1. Start with Mock Mode - Ensures reliable demo without API failures
2. Use Quick Prompts - Pre-configured questions for best results
3. Expand Reasoning - Click to show detailed AI reasoning
4. Compare Modes - Switch between Council, DxO, and Adversarial to show different approaches
5. Select Models - Choose from multiple AI providers (OpenAI, Anthropic, Google, OpenRouter)
6. Adversarial Turns - Configure debate cycles for deeper analysis

License

MIT