MemoryOS

MemoryOS is a contextual AI-powered personal wiki built on top of HydraDB.

Unlike traditional bookmarking or note-taking systems that only store links or content, MemoryOS creates a living semantic memory graph around the user’s knowledge, intent, research journey, and evolving interests.

The project explores a core question:

What if your personal wiki could remember not just what you saved, but why it mattered to you?

Users can save:

research papers tweets YouTube videos articles Reddit posts UI inspirations highlighted text technical resources ideas from anywhere on the internet

through both a web application and a browser extension assistant.

When content is saved, MemoryOS captures:

the user’s reason for saving it highlighted sections nearby contextual text semantic topics workflows entities emotional/project context

The system then:

Scrapes and processes the content Extracts semantic entities and relationships Builds graph-based memory packets Stores them inside HydraDB as contextual long-term memory

Over time, this forms a personalized AI wiki and semantic knowledge graph that reflects:

projects research directions recurring concepts connected ideas long-term interests

Instead of acting like a chatbot or simple vector search system, MemoryOS behaves like an evolving contextual memory layer.

Users can ask questions such as:

“What patterns exist across the AI memory systems I’ve researched?” “Show me memories related to semantic retrieval and long-term AI memory.” “What design themes appear across the UI inspirations I saved?” “What older memories are connected to my robotics research?”

MemoryOS retrieves actual connected memories through HydraDB graph traversal and semantic recall.

The interface presents:

conversational recall narration interactive memory cards semantic anchors graph relationships connected memory clusters contextual evidence long-term knowledge continuity

Key Features:

AI-powered personal wiki generation Semantic memory graph construction Context-aware browser assistant HydraDB graph-based recall Interactive memory exploration Long-term contextual retrieval Semantic clustering and relationship extraction Cross-platform knowledge capture

Tech Stack:

HydraDB → semantic memory engine Supabase → auth + metadata Firecrawl → web scraping NVIDIA NIM + Groq → semantic extraction and reasoning React + TypeScript → interface Chrome Extension → contextual capture assistant

MemoryOS is designed for:

researchers developers founders designers AI engineers students knowledge workers

who constantly collect information online but lose the context behind why it mattered.

The project turns fragmented internet knowledge into a structured, evolving, AI-powered personal wiki.