🧠 Rewind - Your Memory Never Forgets

Proactive memory intelligence that tells you what to do next.

Rewind is built on top of Supermemory API to provide an intelligent layer that doesn't just store your memories—it tells you what to do with them.

🔗 Live Demo: https://rewind-mu.vercel.app

✍️ My Contribution

This project was built as part of a hackathon team. Below are the components I personally designed and implemented:

• Retention Scoring System

  • Designed and implemented the retention scoring formula to quantify memory decay
  • Built rolling 7-day and 30-day retention windows to distinguish short-term vs long-term forgetting patterns

• Dashboard Chatbot

  • Designed the chatbot prompt and interaction logic for user-facing insights
  • Integrated Pinecone-based semantic memory retrieval to surface relevant historical context during conversations

• Learning Pattern Analysis

  • Implemented logic to detect learning and forgetting trends across user activity
  • Enabled proactive reminders and recommendations based on retention signals

• Evaluation & Iteration

  • Validated retention outputs using simulated user timelines
  • Iterated on scoring thresholds to improve stability and interpretability

Collaboration: Built alongside a teammate during a time-boxed hackathon environment, with clear ownership over the above components.

🖥️ Product Walkthrough

📊 Dashboard Overview

High-level dashboard showing memory stats, retention signals, activity trends, and proactive recommendations.

---

🧠 Memory Management

Centralized memory view with categorization, filtering, review status, and time-based organization.

---

🔍 Semantic Search

Natural-language semantic search over stored memories using vector-based retrieval.

---

💬 Ask Recall (AI Chat)

Conversational AI interface that answers user questions by retrieving and reasoning over relevant past memories.

🎯 The Problem

• Supermemory Chat: You ask → it answers (reactive)
• Rewind: It TELLS you what you're forgetting (proactive)

✨ Key Features

🔴 Smart Alerts

• "You have an interview in 3 days. Last time you prepped 5 days before. Start now?"
• Automatic deadline tracking with urgency ranking
• Unreviewed content detection
• NEW: Email digests with daily/weekly summaries

📊 Enhanced Pattern Detection

• Discovers your best study times correlated with quiz performance
• Identifies gaps in your knowledge automatically
• Tracks consistency with streak detection
• NEW: Retention trend analysis (improving/declining/stable)
• NEW: Knowledge gap detection per subject
• NEW: Review habit tracking

🧠 Quiz Mode

• NEW: AI-generated questions from your memories
• NEW: Retention score tracking per memory
• NEW: Subject-based quizzes
• NEW: Self-assessment with detailed results

💬 Chat Auto-Save

• NEW: Conversations automatically become memories
• NEW: Smart detection of learning content
• NEW: Metadata extraction (type, subject, priority)

📝 Memory Templates

• NEW: 6 pre-built templates (Study, Interview, Algorithm, Meeting, Concept, Tip)
• NEW: First-person writing style enforcement
• NEW: Structured metadata extraction

🎯 Context-Aware Modes

• Study Mode: Quiz tracking, spaced repetition, weak topic detection
• Job Search Mode: Interview pipeline, company research, prep tracking
• Meeting Mode: Action items, context from past meetings

⏱️ Timeline Intelligence

• Visual timeline of when you learned something
• When you last reviewed it
• When you should review it again

🚀 Tech Stack

• Frontend: Next.js 14 + TypeScript + Tailwind CSS
• Memory Engine: Supermemory API
• Fonts: Space Grotesk, Syne, Inter
• Icons: Lucide React
• Charts: Recharts
• Date Utils: date-fns

📦 Installation

# Install dependencies
npm install

# Set up environment variables
cp .env.example .env.local
# Add your SUPERMEMORY_API_KEY to .env.local

# Run development server
npm run dev

Open http://localhost:3000 to see the landing page.

🎨 Project Structure

memoryos/
├── app/
│   ├── landing/          # Landing page
│   ├── dashboard/        # Main dashboard
│   ├── api/
│   │   ├── memories/     # Memory CRUD operations
│   │   ├── search/       # Search endpoint
│   │   └── dashboard/    # Dashboard stats & patterns
├── components/
│   └── dashboard/        # Dashboard UI components
├── lib/
│   ├── supermemory.ts    # Supermemory client
│   └── patterns.ts       # Pattern detection engine
└── types/
    └── index.ts          # TypeScript types

🧪 Pattern Detection Engine

The pattern detector analyzes your memories to provide:

1. Urgent Alerts: Approaching deadlines, unreviewed high-priority items
2. Attention Alerts: Knowledge gaps, interview prep imbalances
3. On Track Alerts: Consistency streaks, momentum indicators
4. Behavioral Patterns: Best study times, retention rates, activity trends

🔑 Environment Variables

SUPERMEMORY_API_KEY=your_api_key_here
NEXT_PUBLIC_USER_ID=user_123
GROQ_API_KEY=your_groq_api_key_here  # For AI features (chat, quiz)

# Optional - for email sending
SENDGRID_API_KEY=your_sendgrid_key
# or
RESEND_API_KEY=your_resend_key

Get Groq API key (free): https://console.groq.com/

📝 Usage

Adding Memories

const memory = await fetch('/api/memories', {
  method: 'POST',
  body: JSON.stringify({
    content: "Studied k-means clustering algorithm",
    metadata: {
      type: 'study',
      subject: 'Machine Learning',
      priority: 'high',
      deadline: '2025-11-05T10:00:00Z'
    }
  })
});

Searching

const results = await fetch('/api/search', {
  method: 'POST',
  body: JSON.stringify({
    query: 'clustering algorithms',
    limit: 10
  })
});

🙏 Acknowledgments

Built with Supermemory API - the memory infrastructure for AI apps.