Journal
Blog
Research and product notes while we ship. Honest numbers. No vapor.
RSS
New·Research
Is memory optional for AI agents, or mandatory?
95% of enterprise AI pilots showed no profit impact, and bigger context windows don't fix it. The research, our own test (18 vs 41 of 50), and where the line sits.
Enkeon
Oct 23, 2026 · 9 min read
Research
LongMemEval-S: measuring memory retrieval
An isolated retrieval evaluation: 99.0% Recall@15 on 499 scored questions, with 481 mean context tokens. Method, category results and limitations.
Enkeon
Jul 28, 2026 · 8 min read
Featured·Research
RotBench: the benchmark that catches AI memory rotting
A longitudinal internal evaluation of stale knowledge, corrections and poisoned tool output, including failures found in our own engine.
Enkeon
Jul 22, 2026 · 8 min read
Featured·Engineering
CompAG: Compile-Augmented Generation
An engineering account of compiling sources into structured knowledge before retrieval, including source evidence and tradeoffs.
Enkeon
Jun 30, 2026 · 6 min read