AriadneMemory for AI agents

Sub-millisecond search. Zero infrastructure.

Get Started

GitHub

0.89ms vector search

FAISS-powered. 12× faster than sqlite-vec. Scales to millions of memories.

Hybrid retrieval

Vector similarity + keyword search + graph traversal. Reciprocal Rank Fusion. 90%+ recall (with semantic embeddings).

Knowledge graph

Typed entities, relationships, multi-hop traversal. One query walks the full chain.

Cognitive retention

Ebbinghaus forgetting curve. Memories strengthen with use, fade without it.

Auto-dedup

MinHash LSH catches near-duplicates at 1.25ms before they enter the system.

Zero infrastructure

SQLite + FAISS. One .db file. No Docker, no Redis, no API keys, no daemon.

Quick Start

python

from arriadne import AriadneMemory

memory = AriadneMemory("./my-memory.db")

# Store a memory
memory.add(
    content="VPS has 4 cores, 8GB RAM, Ubuntu 24.04",
    source="system",
    importance=0.8,
)

# Search — vector + keyword + graph, fused
results = memory.search("server specs", limit=5)
# → [<Memory content="VPS has 4 cores..." score=0.94>]

	Ariadne	Mnemosyne	Mem0	ChromaDB
Vector search	0.89ms	153ms	12ms	8ms
Hybrid search	✅	❌	❌	⚠️
Knowledge graph	✅	⚠️	❌	❌
Runs locally	✅	✅	❌	✅
No daemon	✅	✅	❌	❌

bash

pip install arriadne