evaluation
The 10 Things to Look For in an Agent-Memory System
The 10 things that separate a real agent-memory system from a demo: retrieval, conflict resolution, freshness, security, self-hosting, observability, and more.
Hindsight Is #1 on BEAM — the Benchmark That Tests Memory at 10 Million Tokens
Hindsight is #1 on BEAM — the memory benchmark that tests at 10M tokens where context stuffing is impossible. See every published result and what drives the 58% margin.
Agent Memory Benchmark: A Manifesto
Agent Memory Benchmark: A Manifesto