Evaluation ReportActive
The LoCoMo Benchmark Report
Our core evaluation matrix checking long-context recall, entity traversal, and temporal consistency across 1,540 simulated developer query runs.
ACCURACY: 70.5% | HIT@10: 79.9%Read Study
Memwyre Research Lab
Advancing AI Long-Term Memory
Peer-reviewed benchmark datasets, architecture studies, and evaluations exploring hierarchical entity graphs, context compression, and stateful agent systems.
$npx -y install-memwyre
LoCoMo:70.5% Accuracy · 79.9% Hit@10
1,540 Test Evaluations
Open Source Reproducibility
Architecture Study6 min read
What is AI Memory? The Architecture of Long-Term Context
Explore the technical architecture of AI memory, from flat vector embeddings to hierarchical entity graphs, and how it solves LLM amnesia.
Apr 2026Read Study
Architecture Study16 min read
Vector Database vs. AI Agent Memory
The fundamental engineering differences between passive semantic vector tables and active cognitive agent memory layers with Ebbinghaus decay.
Sep 2026Read Study
Developer Guide7 min read
Context Engineering: The 2026 Guide for Developers
Prompt engineering is dead. Learn how Context Engineering and hierarchical entity graphs replace standard RAG for production AI agents.
May 2026Read Study
The second brain
for your AI.
Access open benchmarks, stateful context graphs, and research datasets. Set up in under 2 minutes.
