Blog

Notes on agent memory

Thinking out loud about memory, context, and consistency - and what we're learning while building Jennah with our first design partners.

Beyond the memory layer: why we built a globally-consistent backend for AI agents

Memory is only a fraction of the problem. Production agent stacks are a distributed-systems nightmare of vector stores, relational data, graphs, and homegrown orchestration. Here's how we unified memory, application data, and durable execution into one consistent backend.

Read post →

Why agent memory shouldn't be a stack of databases

Most teams glue a vector database to a graph to a log store and spend their time keeping the three in sync. Here's why we think agent memory belongs in one consistent store - and what that unlocks.

Read post →