AI-Powered Product Management Knowledge Hub
ProductSnap is a comprehensive content aggregator designed for product managers, providing curated articles from 86 RSS feeds, 298 Lenny's Podcast transcripts, and an AI-powered chat assistant with full RAG search across all content.
- 86 Curated RSS Feeds - Vetted, healthy feeds from top PM publications, thought leaders, and tech companies
- 298 Lenny's Podcast Transcripts - Full searchable transcripts (4.5M+ words) from the #1 PM podcast
- 3,300+ Articles - Continuously growing knowledge base
- Smart Feed Refresh - Priority-based updates every 4 hours with health scoring
- My Files - Upload your own documents (PDF, TXT, MD, DOCX) and chat with them
- Full Knowledge Base Search - AI searches ALL articles, podcasts, and your files
- Multi-provider Support - OpenAI (GPT-4o, o1), Anthropic (Claude), Google (Gemini)
- Token-Optimized Context - Tiered snippet strategy saves ~47% tokens while keeping all 50 sources
- Word-Boundary Matching - Precise
\bregex matching eliminates false positives from substring matches - Improved Relevance - Query intent analysis, minimum score thresholds, normalized scoring
- Collapsible Sources - Expand/collapse sources with relevance indicators (high/medium/low)
- Rich Formatting - Syntax-highlighted code blocks, tables, lists, and blockquotes
- Source Citations - Every AI response includes clickable source references
- Copy Code Button - One-click copy for code snippets with language labels
- Bookmarks - Save articles and podcasts with notes, export to Notion/Obsidian
- Read History - Track what you've read with duration tracking
- My Analytics - Personal reading stats, top categories, search history
- My Files - Upload documents to include in AI chat context
- AI Summaries - Generate article/podcast summaries on demand
- Email Digest - Daily/weekly email digest subscription (configurable)
- Feed Health Scores - 0-100 health rating based on activity and errors
- Priority Refresh - Active feeds refreshed more frequently
- Error Tracking - Consecutive errors and automatic health degradation
- Admin Dashboard - Feed health overview with healthy/warning/unhealthy counts
- Google OAuth - Secure login with Google accounts
- Role-based Access - Admin and User roles with different permissions
- AES-256-GCM Encryption - Authenticated encryption for stored API keys
- httpOnly Cookies - XSS-safe token storage
- Rate Limiting - Protection against abuse on all endpoints
- SSRF Protection - URL validation on content extraction
- Server-Side Pagination - Fetches 24 items per page instead of loading entire database
- Memory-Efficient Caching - Article content stripped from cache, loaded on demand (~28MB saved)
- Lazy-Loaded Transcripts - Podcast transcripts loaded from disk on demand (~48MB saved)
- Analytics Pruning - Automatic 30-day retention prevents unbounded database growth
- Debounced Search - 300ms debounce prevents excessive API calls while typing
- Non-Blocking DB Writes - Async retry logic prevents event loop blocking
- Apple-Inspired Design - Clean, minimal aesthetic with smooth animations
- Light & Dark Themes - System-aware theme switching
- Glass Morphism - Frosted glass header with backdrop blur
- In-App Reader - Read articles without leaving the app
- Transcript Viewer - Full-text podcast transcripts with word-boundary search
- Responsive Design - Works beautifully on desktop, tablet, and mobile
- Node.js >= 18.0.0
- npm >= 9.0.0
- Google Cloud OAuth credentials
# Clone the repository
git clone https://github.com/yourusername/productsnap.git
cd productsnap
# Install all dependencies
npm run setup
# Or manually:
npm install
cd client && npm install && npm run build
cd ..- Create
.envfile in the root directory:
# Google OAuth Configuration
GOOGLE_CLIENT_ID=your_google_client_id
GOOGLE_CLIENT_SECRET=your_google_client_secret
GOOGLE_CALLBACK_URL=http://localhost:3000/api/auth/google/callback
# JWT Configuration (32+ characters required)
JWT_SECRET=your_32_char_secret_key_here
# Encryption Key for API Keys (32+ characters required)
ENCRYPTION_KEY=your_32_char_encryption_key_here
# Admin Configuration
ADMIN_EMAIL=your_admin_email@gmail.com
# Server Configuration
PORT=3000
NODE_ENV=development
# Session Secret (32+ characters required)
SESSION_SECRET=your_session_secret_key_here
# Production only
FRONTEND_URL=https://yourdomain.com- Set up Google OAuth:
- Go to Google Cloud Console
- Create a new project or select existing
- Enable Google+ API
- Go to Credentials -> Create OAuth Client ID
- Set authorized redirect URI:
http://localhost:3000/api/auth/google/callback - Copy Client ID and Secret to
.env
# Start the server
npm start
# Or for development
npm run devVisit http://localhost:3000 in your browser.
Backend:
- Express.js - Web framework
- LowDB - JSON file database
- Passport.js - Google OAuth authentication
- JSON Web Tokens - Session management (httpOnly cookies)
- Node.js crypto - AES-256-GCM encryption
- rss-parser - RSS feed parsing
- multer - File upload handling
- pdf-parse - PDF text extraction
- adm-zip - DOCX file parsing
- OpenAI/Anthropic/Google SDKs - AI providers
Frontend:
- React 19 - UI framework
- React Router - Client-side routing
- Tailwind CSS - Styling
- Radix UI - Accessible components
- Lucide React - Icons
- react-markdown - Markdown rendering
- rehype-highlight - Syntax highlighting for code blocks
productsnap/
├── aggregator-server.js # Main Express server
├── package.json
├── .env # Environment variables (not in git)
├── content-aggregator.json # Database file
│
├── middleware/
│ ├── auth.js # JWT authentication (httpOnly cookie)
│ └── rbac.js # Role-based access control
│
├── routes/
│ ├── auth.js # Google OAuth routes
│ ├── settings.js # User settings API
│ ├── chat.js # AI chat endpoint with RAG
│ └── admin.js # Admin management
│
├── services/
│ ├── encryption.js # AES-256-GCM encryption
│ ├── ai/
│ │ ├── index.js # AI service factory
│ │ ├── openai.js # OpenAI integration
│ │ ├── anthropic.js # Anthropic integration
│ │ └── google.js # Google AI integration
│ └── rag/
│ └── search.js # RAG content search (50 sources)
│
├── uploads/ # User uploaded files (per-user folders)
│
├── client/ # React frontend
│ ├── src/
│ │ ├── App.jsx # Main app component
│ │ ├── main.jsx # Entry point
│ │ ├── index.css # Apple-inspired themes
│ │ ├── context/
│ │ │ └── AuthContext.jsx
│ │ └── components/
│ │ ├── ui/ # ShadCN/Radix UI components
│ │ ├── auth/ # Login, UserMenu
│ │ ├── chat/ # ChatBox, ChatMessage (with collapsible sources)
│ │ ├── settings/ # SettingsPage
│ │ ├── admin/ # AdminPanel, UserManager
│ │ ├── bookmarks/ # BookmarksPage
│ │ ├── analytics/ # AnalyticsPage
│ │ └── files/ # MyFilesPage
│ └── dist/ # Built frontend
│
└── Lenny's Podcast Transcripts Archive/
└── *.txt # 298 transcript files
| Endpoint | Method | Description |
|---|---|---|
/api/articles |
GET | Get paginated articles |
/api/articles/:id |
GET | Get single article |
/api/podcasts |
GET | Get podcast transcripts |
/api/podcasts/:id |
GET | Get single transcript |
/api/search |
GET | Search articles & podcasts |
/api/categories |
GET | Get all categories |
/api/feeds |
GET | Get feed statistics |
/api/stats |
GET | Get overall statistics |
| Endpoint | Method | Role | Description |
|---|---|---|---|
/api/auth/me |
GET | Any | Get current user |
/api/auth/logout |
POST | Any | Logout |
/api/settings |
GET/PUT | Any | User preferences |
/api/settings/api-keys |
PUT | Any | Save API keys |
/api/chat |
POST | Any | AI chat with RAG (50 sources) |
/api/chat/providers |
GET | Any | Get AI providers |
/api/bookmarks |
GET/POST | Any | List/add bookmarks |
/api/bookmarks/:id |
DELETE/PATCH | Any | Remove/update bookmark |
/api/history |
GET/POST | Any | Get/track read history |
/api/history/check |
GET | Any | Check if item is read |
/api/analytics |
GET | Any | Get user analytics |
/api/summarize |
POST | Any | Generate AI summary |
/api/files |
GET/POST | Any | List/upload files |
/api/files/:id |
GET/DELETE | Any | Get/delete file |
/api/export/bookmarks |
GET | Any | Export bookmarks (JSON/MD/CSV) |
/api/export/history |
GET | Any | Export read history |
/api/digest/subscribe |
POST/DELETE | Any | Subscribe/unsubscribe digest |
/api/digest/status |
GET | Any | Get digest subscription status |
/api/digest/preview |
GET | Any | Preview digest content |
/api/feeds/health |
GET | Admin | Get feed health status |
/api/admin/users |
GET | Admin | List all users |
/api/admin/users/:id/role |
PUT | Admin | Change user role |
/api/refresh |
POST | Admin | Manual feed refresh |
The AI chat searches ALL 3,300+ articles, 298 podcast transcripts, and your uploaded files to find the most relevant content for your question.
- Your question is analyzed for keywords, phrases, and intent (who/what/how questions)
- Every article, podcast, and your uploaded files are scored for relevance
- Precise relevance filtering:
- Word-boundary regex matching (
\b) eliminates false positive substring matches - Minimum score thresholds (relevance: 5, normalized: 8) filter noise
- Normalized scoring for fair comparison across content lengths
- Query intent analysis boosts relevant content types
- Quoted terms and proper nouns get priority matching
- Word-boundary regex matching (
- Top 50 most relevant sources are selected
- Tiered snippets are extracted to optimize token usage:
- Tier 1 (Top 10): 800-char snippets for highest relevance
- Tier 2 (Next 15): 400-char snippets for good coverage
- Tier 3 (Last 25): 150-char snippets for breadth
- Transcript content is loaded lazily from disk (not held in memory)
- AI generates a response with rich markdown formatting
- Sources are cited with relevance indicators
- Collapsible Sources - Expand/collapse source panel, shows relevance (high/medium/low)
- Syntax Highlighting - Code blocks with language detection and GitHub Dark theme
- Copy Button - One-click copy for code snippets
- Tables - Formatted comparison tables
- Blockquotes - Styled quotes from sources
- Lists - Bullet and numbered lists with proper spacing
- Your Files - Uploaded documents are automatically included in search
- Login with Google account
- Go to Settings -> API Keys
- Add your API key for OpenAI, Anthropic, or Google AI
- Select your preferred model in Settings -> Preferences
- Click the chat icon in the header
- Ask questions like:
- "What does Brian Chesky say about product management?"
- "Summarize best practices for product launches"
- "What are the key metrics for product-led growth?"
- "Compare what different guests said about hiring PMs"
OpenAI:
- GPT-4o (Latest)
- GPT-4o Mini
- GPT-4 Turbo
- o1, o1-mini, o1-preview
Anthropic:
- Claude Sonnet 4
- Claude Opus 4
- Claude 3.5 Sonnet/Haiku
- Claude 3 Opus/Sonnet/Haiku
Google:
- Gemini 1.5 Pro
- Gemini 1.5 Flash
- Gemini 2.0 Flash (Experimental)
Upload your own documents and they'll be included in AI chat searches automatically.
- TXT - Plain text files
- PDF - PDF documents (text extracted automatically)
- MD - Markdown files
- DOCX - Microsoft Word documents
- Go to My Files (via navigation or user menu)
- Upload documents (max 10MB each, up to 5 at once)
- Text content is extracted and stored securely
- When you chat with AI, your files are searched alongside articles and podcasts
- Your files get a relevance boost to prioritize personal content
- Files are stored per-user in isolated directories
- Only you can see and search your uploaded files
- Delete files anytime from the My Files page
Save articles and podcasts to read later:
- Add notes to bookmarks
- Export to JSON, Markdown, or Notion CSV
- Filter and search your bookmarks
Track your reading activity:
- Automatic tracking when you view content
- Read count and total time tracking
- Export your reading history
View your personal stats:
- Articles and podcasts read
- Search and chat activity
- Top categories you read
- Recent search history
- Admin users see system-wide analytics
Generate concise summaries of any article or podcast:
- 3-5 bullet point summaries
- Uses your configured AI provider
- Summaries are cached to avoid regeneration
Subscribe to periodic content updates:
- Daily or weekly frequency
- Filter by preferred categories
- Preview digest before subscribing
ProductSnap implements comprehensive security measures:
- No Hardcoded Secrets - All secrets must be set via environment variables
- AES-256-GCM Encryption - API keys encrypted with random IV and authentication tag
- httpOnly Cookies - JWT tokens stored in httpOnly cookies (XSS-safe)
- SameSite Cookies - CSRF protection via strict same-site policy
- SSRF Protection - URL validation blocks internal network access
- Input Validation - Length limits and format validation on all inputs
- Rate Limiting - Per-endpoint rate limits prevent abuse
- ReDoS Protection - Safe string matching instead of user-controlled regex
# Start backend
npm run dev
# In another terminal, start frontend with hot reload
cd client
npm run dev# Build frontend
npm run build:client
# Start production server
NODE_ENV=production npm start-
Create Railway Account at railway.app
-
Connect GitHub Repository
- Click "New Project" -> "Deploy from GitHub repo"
- Select your ProductSnap repository
-
Add Environment Variables in Railway Dashboard:
NODE_ENV=production GOOGLE_CLIENT_ID=your_google_client_id GOOGLE_CLIENT_SECRET=your_google_client_secret GOOGLE_CALLBACK_URL=https://your-app.railway.app/api/auth/google/callback JWT_SECRET=your_32_char_secret_key ENCRYPTION_KEY=your_32_char_encryption_key SESSION_SECRET=your_session_secret ADMIN_EMAIL=your_admin_email@gmail.com FRONTEND_URL=https://your-app.railway.app -
Update Google OAuth - Add Railway URL to authorized redirect URIs in Google Cloud Console
-
Deploy - Railway will automatically build and deploy
The app is optimized to run within 512MB memory on Railway:
- Article content is stripped from the in-memory cache (~28MB saved)
- Podcast transcripts are loaded from disk on demand (~48MB saved)
- Analytics data is pruned to 30-day retention on startup and daily at 6 AM
- Feed refresh runs every 4 hours to reduce memory churn
- Cache TTL is enforced at 60 seconds to prevent stale data buildup
If you see UNKNOWN: unknown error, open 'content-aggregator.json':
- The app has async retry logic (
safeDbWrite) for OneDrive sync conflicts - Non-blocking retries with exponential backoff (200ms, 400ms, 600ms)
- Concurrency is reduced to 3 parallel feed fetches
- Wait for OneDrive to finish syncing before restarting
- Ensure redirect URI matches exactly in Google Cloud Console
- Check that GOOGLE_CLIENT_ID and GOOGLE_CLIENT_SECRET are correct
- Verify GOOGLE_CALLBACK_URL matches your server URL
- Verify API key is correct in Settings
- Check that you've selected a model for the provider
- Look at server logs for API error messages
- Fork the repository
- Create a feature branch (
git checkout -b feature/amazing-feature) - Commit your changes (
git commit -m 'Add amazing feature') - Push to the branch (
git push origin feature/amazing-feature) - Open a Pull Request
This project is licensed under the MIT License.
- Lenny Rachitsky for the amazing podcast transcripts
- All the PM thought leaders whose content powers this aggregator
- The open-source community for the amazing tools and libraries
Built with love for the Product Management community