Skip to main contentWhat Is CORE?
CORE is a memory agent that replicates how human memory actually works. Your brain doesn’t store memories as plain text—it organizes episodes into topics, creates associations, and knows where things belong. CORE does the same for your AI tools.
The Core Concepts
How your knowledge is structured: Episodes, Entities, Statements
The 11 categories that classify every fact (Preference, Decision, Directive, etc.)
The 5 ways CORE understands what you’re looking for (Aspect, Entity, Temporal, etc.)
How people, projects, technologies, and concepts are categorized
How episodes are automatically organized by topic
How It Works in 30 Seconds
- You talk to Claude/Cursor/ChatGPT → CORE captures the conversation
- Facts are extracted → “Manik prefers TypeScript” becomes a structured statement
- Categorized by aspect → This is a Preference (1 of 11 types)
- Stored in knowledge graph → Connected to entities (Manik, TypeScript)
- Surfaced when needed → Next time Claude writes code, your Preference statements are available
Memory Agent vs RAG
Traditional RAG: “Search all text for similarity”
CORE: “Understand what you’re asking, search precisely”
When you ask “What are my coding preferences?”:
- RAG: Searches everything, returns 50 chunks, hopes relevant ones are there
- CORE: Classifies as Aspect Query, filters by
aspect=Preference, returns 5 precise facts
Result: 3-4x faster, infinitely more accurate.
Next Steps
