Revolutionary neural memory format inspired by biological memory traces. Built for the AI era with breakthrough performance and intelligent architecture.
Engram draws inspiration from neuroscience research on memory engrams — the biological traces of memory storage in the brain. Just as neurons form hierarchical networks with temporal dynamics, Engram creates intelligent memory structures that understand context and time.
Hierarchical Navigable Small World graphs enable logarithmic search complexity, making real-time semantic search practical at massive scale.
Like biological memory, Engram gives higher weight to recent memories while preserving important historical context through intelligent decay functions.
Store and search across text, images, audio, and structured data in a unified neural memory space with cross-modal semantic understanding.
Memory consolidation, forgetting curves, and synaptic strengthening patterns inform Engram's architecture for more human-like intelligence.
From linear O(n) brute force to logarithmic O(log n) graph traversal — a fundamental breakthrough that makes real-time AI memory practical.
400x performance improvement validated across 1000+ node datasets
Get up and running with Engram in minutes. Available on NPM with TypeScript support and comprehensive documentation.
Engram represents the culmination of years of research into AI memory systems, building on the foundation of AIF-BIN while achieving breakthrough performance through neural inspiration.