2026-07-29
I finished the bi-temporal part of Jennah’s episodic memory over the weekend; basically time-travel queries. I needed it so I could run LongMemEval honestly against it. Asking “what did the user believe back in March?” is very different from “what is true now?”, and if the memory system can’t distinguish between the two, it just ends up guessing. For the benchmark setup, I used Claude Opus 4.8 as both the judge and answer generator, with gemini-embedding-001 for vector embeddings (Jennah supports managed embedding as well as BYO-vectors). I ran all 500 questions from longmemeval_s_cleaned.json.
Agent · Agent · Ai · Ai · Benchmark · Benchmark · Jennah · Jennah · Memory · Memory
2 minutes
A little something I’ve been building
2026-07-12
I’ve been spending a lot of my nights and weekends on Jennah. It kind of nagged its way into existence: AI agents are getting genuinely good, but the memory underneath them is still held together with tape. Everyone builds it the same way: a vector database for recall, a separate graph database for reasoning, an execution log off in the corner, and a pile of app-side glue trying to keep the three in sync. That’s the thing I wanted to address. Jennah’s whole idea is one store; execution logs, vector recall, and reasoning graphs living together in a single store per agent, instead of a vector DB bolted onto a graph DB bolted onto a log.
Agent · Agent · Ai · Ai · Context · Context · Jennah · Jennah · Memory · Memory
2 minutes