Engram: A Local-First Persistent Memory Architecture for Conversational AI Agents

Zenodo (CERN European Organization for Nuclear Research) 2026 Open access

Read full text
Authors: Cartisien
DOI: 10.5281/zenodo.18988891
Cited by: 0

Abstract

Engram is a local-first persistent memory system for AI agents combining multi-strategy retrieval (semantic embeddings, BM25, temporal decay via Reciprocal Rank Fusion), a three-tier memory lifecycle with LLM-based consolidation, a sparse entity-relationship graph, and a belief-revision model. Runs on consumer hardware with no cloud dependency. Ships TypeScript and Python SDKs and an MCP server for Claude Desktop/Cursor integration.

Fields of Study & Subjects

Information Retrieval and Search Behavior Topic Modeling Artificial Intelligence Applications

Publication details

Journal / source
Zenodo (CERN European Organization for Nuclear Research)
Publication date
12 March 2026
Language
EN