Skip to content

SimranShaikh20/MindMesh-AI

Repository files navigation

MindMesh AI - Multi-Agent Research System

MindMesh AI Gemini Status

MindMesh AI is an intelligent research assistant using 6 specialized AI agents working in parallel to analyze complex questions from multiple perspectives and provide balanced, actionable recommendations in real-time.

🎯 Live Demo: [https://mind-mesh-ai-two.vercel.app/]


🌟 Key Features

  • 🚀 Ultra-Fast Parallel Processing: All agents analyze simultaneously using Google Gemini AI
  • 🎭 6 Specialized AI Agents: Research, Pro/Con Advocacy, Bias Checking, Fact Verification, Synthesis
  • ⚡ Real-Time Streaming: Watch agents "think" and respond live with streaming updates
  • 🎨 Beautiful Modern UI: Dark theme with purple-pink gradients and smooth animations
  • 📱 Fully Responsive: Works seamlessly on mobile, tablet, and desktop
  • 📊 Confidence Meter: Visual confidence scoring with animated progress bar
  • 📜 History Sidebar: Track and revisit your past 10 analyses
  • 💡 Smart Follow-ups: AI-generated contextual follow-up questions
  • ⚡ Speed Banner: Real-time comparison of parallel vs sequential processing

🤖 The 6 AI Agents

Each agent has a distinct personality and purpose:

Phase 1: Initial Analysis (Parallel)

  1. 📊 Research Agent (Gemini Flash)

    • Role: Data & Statistics Analyst
    • Focus: Objective facts, trends, statistics
  2. 💡 Pro Advocate (Gemini Flash)

    • Role: Opportunity Finder
    • Focus: Benefits, opportunities, positive outcomes
  3. 😈 Con Advocate (Gemini Flash)

    • Role: Risk Assessor
    • Focus: Risks, downsides, potential problems

Phase 2: Quality Assurance (Parallel)

  1. 🎯 Bias Checker (Gemini Pro)

    • Role: Critical Analyst
    • Focus: Identifies logical fallacies and cognitive biases
  2. ✅ Fact Checker (Gemini Pro)

    • Role: Truth Verifier
    • Focus: Verifies claims and flags misinformation

Phase 3: Synthesis

  1. 🎓 Synthesizer (Gemini Pro)
    • Role: Strategic Advisor
    • Focus: Creates balanced final recommendation

🚀 Tech Stack

  • Frontend: React 18, TypeScript, Vite, Tailwind CSS
  • Backend: Lovable Cloud (Supabase) Edge Functions
  • AI Models:
    • Google Gemini 2.5 Flash (fast agents)
    • Google Gemini 2.5 Pro (deep reasoning agents)
  • Deployment: Lovable Platform (serverless, auto-scaling)

💻 Architecture

User Query
    ↓
Edge Function (TypeScript)
    ↓
Phase 1: Research, Pro, Con → PARALLEL (Gemini Flash)
    ↓
Phase 2: Bias, Fact Checker → PARALLEL (Gemini Pro)
    ↓
Phase 3: Synthesizer → Final Recommendation (Gemini Pro)
    ↓
Streaming Response → Frontend
    ↓
Beautiful UI Display

Processing Time: ~3-5 seconds for complete analysis (vs 20+ seconds if sequential)


🎯 Example Questions

Try asking:

  • "Should I switch careers to AI/ML engineering?"
  • "Is buying a house in 2025 a good financial decision?"
  • "Should I start a SaaS business or get a job?"
  • "Is remote work better than office work?"

🏆 Hackathon Alignment

Sponsor Track How MindMesh AI Qualifies
Cerebras Parallel agent execution with ultra-fast inference
Meta Llama Use of Gemini LLM models for fast and deep reasoning agents
Docker (Deployment) Serverless deployment on Lovable Platform with auto-scaling

🛠️ Local Development

Prerequisites

  • Node.js 18+
  • npm or yarn

Setup

  1. Clone the repository
git clone https://github.com/SimranShaikh20/MindMesh-AI
cd mindmesh-ai
  1. Install dependencies
npm install
  1. Run development server
npm run dev
  1. Open browser
http://localhost:8080

This app is already connected to Lovable Cloud with AI enabled - no API keys needed!


🚀 Deployment

This app is deployed on Lovable Platform with:

  • ✅ Automatic deployments
  • ✅ Serverless edge functions
  • ✅ Global CDN
  • ✅ Auto-scaling
  • ✅ SSL certificates

To deploy your own:

  1. Fork this project in Lovable
  2. Click "Publish" in the top right
  3. Your app is live!

🎨 Design System

Color Palette:

  • Background: Dark gradient (gray-900 → purple-900)
  • Primary: Electric purple (#a855f7)
  • Secondary: Hot pink (#ec4899)
  • Cards: Dark slate (gray-800)

Typography:

  • Headers: Bold, gradient text
  • Body: Clean, readable sans-serif
  • Code: Monospace for technical content

Animations:

  • Fade-in entrances
  • Slide-up effects
  • Pulse indicators
  • Staggered agent reveals

📊 Agent Response Format

Standard Agents (Research, Pro, Con, Bias, Fact)

  • 3-4 concise sentences
  • Personality-driven perspective
  • Evidence-based reasoning

Synthesizer

  • Recommendation: Clear decision
  • Key Reasoning: 2-3 bullet points
  • Action Steps: Numbered list (3-5 steps)
  • Confidence Level: X/10 with justification
  • Important Considerations: Risks and caveats

🔒 Security & Privacy

  • ✅ No user data stored
  • ✅ Secure edge functions
  • ✅ Rate-limited AI requests
  • ✅ Input validation
  • ✅ CORS protection

💡 Why MindMesh AI?

Traditional decision-making often suffers from:

  • Confirmation bias: We seek information that confirms existing beliefs
  • Overconfidence: We overestimate our knowledge
  • Lack of diverse perspectives: We miss important viewpoints

MindMesh AI solves this by:

  • Forcing consideration of pro AND con arguments
  • Using dedicated agents to check for biases
  • Verifying factual claims
  • Synthesizing a balanced recommendation

🌟 Key Innovations

  1. Parallel Agent Processing: 3-5x faster than sequential
  2. Real-Time Streaming: Live agent updates
  3. Bias Detection: Dedicated quality assurance agents
  4. Beautiful UX: Professional, modern interface
  5. Confidence Meter: Extract and display confidence scores (X/10) with animated color-coded bars
  6. History Sidebar: Collapsible sidebar storing last 10 analyses in localStorage
  7. Smart Follow-ups: AI-generated follow-up questions displayed as clickable gradient cards
  8. Speed Banner: Animated comparison showing parallel vs sequential processing time

📈 Performance

  • Initial Load: < 2 seconds
  • Agent Processing: 3-5 seconds total
  • Streaming Updates: Real-time (< 100ms latency)
  • Mobile Performance: Optimized for all devices

📝 License

MIT License - feel free to use this for your own projects!


About

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Contributors 2

  •  
  •  

Languages