Published May 30, 2026
I was the bottleneck. Every time my AI memory needed maintenance — fixing stale references, deduplicating entries, compressing bloated neurons — I had to be there. With 4 businesses and 38 memories to maintain, the system worked but it depended on me opening a session and running the Librarian manually. That doesn't scale.
We put a local LLM on the machine. Not a cloud API — a model running locally, 24/7, with one job: curate the memory autonomously. No human review loop. No approval queue. The constitution and git history are the controls.
The first iteration showed exactly where the gaps were. Within five hours, the autonomous curator had generated five false positives — flagging connections between unrelated projects because they shared generic tags. The model's JSON parse rate was sitting at around 48%. The system needed work.
We switched to a more capable local model. Parse rate jumped to 100% overnight. We tightened the tag matching threshold, added a blocklist for generic terms, and hardened the deduplication prompts. Simple changes, big impact.
One week later: 140 autonomous cycles, one every hour, zero errors. The system was cross-referencing, deduplicating, and compressing without anyone watching. The moment I realized the architect role — the one I'd spent months building — had made itself nearly obsolete.
EIDARA v2 is open-source and free. What's running on my machine now is something else. A local LLM that curates your memory while you sleep. Many AIs contribute knowledge via MCP. One autonomous curator decides what gets integrated. No cloud, no API keys, no vendor. Just a model running on your hardware, maintaining context that makes every AI session better than the last.
— Javier
EIDARA v2 is free. SUPER DARA is what comes next.