Memory Evolution

CortexDB's knowledge graph is not static. As new episodes are ingested, the graph evolves to reflect current reality while preserving history.

Relationship Updates

When new evidence supports or contradicts existing relationships, the graph updates:

• New relationships are added as entities interact in new ways
• Temporal boundaries are maintained -- relationships track when they started
• Contradictions are detected when new information conflicts with existing relationships

Adaptive Entity Types

CortexDB recognizes a broad set of entity types and adapts to your data:

• Built-in types -- People, organizations, technologies, services, projects, and features are always recognized
• Discovered types -- New entity types are identified from your data patterns as they emerge

Example

Day 1: "payments service uses PostgreSQL"
  → payments_service --uses--> PostgreSQL

Day 30: "migrating payments from PostgreSQL to CockroachDB"
  → payments_service --uses--> CockroachDB
  → payments_service --previously_used--> PostgreSQL

Day 60: "CockroachDB migration complete, all tests passing"
  → payments_service --uses--> CockroachDB (confirmed)

The graph reflects the current state while CortexDB preserves the complete history, so you can always ask "What did we use before?"