BetterDB

BetterDB is the open-core memory and caching layer for AI agents, running on

any Valkey or Redis you own.

Memory gives your agents durable, scoped recall: remember and recall with

ranking by similarity, recency, and importance, so an agent carries context

across sessions instead of starting cold.

Caching answers repeat work fast: a semantic cache catches paraphrased hits, a

multi-tier exact-match cache serves LLM responses, tool results, and session

state in under a millisecond.

The difference from a hosted black box is that you see every decision on a

dashboard and can change how it behaves while the app is running. Thresholds,

TTLs, invalidation, all observable and adjustable live. The cache can even

tune itself: an MCP agent reads live state and proposes a change, a human

approves it, and the running app picks it up in seconds with no redeploy.

Run it self-hosted on your own Valkey, or fully managed. Either way, no vendor

lock-in. MIT core, TypeScript and Python. Live demo at chat.betterdb.com.