Inspiration
In our digital age, we're drowning in personal data - thousands of text messages, emails, PDFs, and documents scattered across our devices. We've all experienced that frustrating moment of knowing we discussed something important with someone, or saved a crucial document, but spending hours trying to find it. Traditional search tools fail because they rely on exact keyword matching, missing the context and meaning behind our conversations. We were inspired to create an AI-powered solution that doesn't just search through your data, but actually understands it, making your digital memories instantly accessible and useful.
What it does
MemorySync is an intelligent personal data assistant that transforms your scattered digital information into a searchable, AI-powered knowledge base. It allows you to interact with your personal data in a natural and intuitive way, making it easy to find the information you need quickly. With Semantic Message Search, you can ask natural language questions like “What did Sarah say about the project deadline?” and get precise results from your iMessage history using advanced vector embeddings. MemorySync understands the context of your messages, so you don’t have to manually sift through hundreds of conversations. The platform also provides Intelligent Conversation Summarization, generating AI-powered summaries of your conversation history with specific contacts. These summaries are organized by time periods, helping you quickly catch up on key discussions without reading every message. Universal Document Retrieval makes finding PDFs and other document attachments effortless. MemorySync uses fuzzy matching to locate relevant files across your message history, so important documents are never lost or hard to find. Privacy is a top priority: all processing happens locally on your machine, ensuring that your personal data never leaves your device. Combined with a natural language interface, MemorySync feels like chatting with a knowledgeable assistant, offering a sleek, modern experience for managing and accessing your digital information.
How we built it
Backend (Python Flask):
AI Integration: Ollama with Llama3 for conversation summarization and natural language understanding
Semantic Search: SentenceTransformers with ChromaDB for vector-based message searching
Database Integration: Direct SQLite integration with macOS Messages database
Fuzzy Matching: RapidFuzz for intelligent PDF filename matching
High-Performance Core (C++):
Custom database operations for lightning-fast message retrieval
Managed with vcpkg for dependency management
CMake build system for cross-platform compatibility
Frontend (React + TypeScript):
Modern, responsive chat interface built with React
Real-time message streaming and typing indicators
File download capabilities for retrieved documents
Custom CSS animations for a smooth user experience
Development Environment: Python virtual environment with comprehensive dependency management Automated build system with CMake RESTful API architecture connecting all components
What's next for MemorySync
Multi-Platform Support: Extend beyond macOS to support Windows and Linux messaging platforms
Enhanced File Types: Add support for images, videos, and other media files with content analysis
Advanced Analytics: Implement relationship mapping and communication pattern analysis
Log in or sign up for Devpost to join the conversation.