Published May 25, 2026
A week ago I turned on the autonomous side of DARA. Here's the update, with the real numbers.
It looks after 38 memories and 14 working agents, and over the past week it ran about 140 times — once an hour, every hour the machine was on. Each cycle is the same quiet round: scan every memory, write a one-line summary for any that lacks one, look for memories that should be cross-linked, clear dead references, flag what's old enough to archive, and rebuild the compiled context the system actually reads from. That context is roughly 7,500 tokens, distilled from about 97,000 tokens of raw memory — a ~13× compression, recomputed every single cycle. This is what agent memory consolidation looks like in practice: persistent memory that stays useful without human curation. Across the whole week: zero errors in the log.
Two things changed along the way, and both are worth being honest about.
The autonomous side runs on a model that lives entirely on my machine — nothing leaves the device. That's the local-first principle EIDARA is built on. The first model was a little too simple for the judgment calls the work needs, so I moved to a more capable one. The field has a name for this kind of component now — sleep-time compute, an agent that does its thinking in the downtime between sessions. That's exactly what DARA's autonomous side is: a sleep-time agent that curates and consolidates memory while you're away.
A diagnostic at the end of the week found a real gap. The system was good at the routine work, but it couldn't tell the difference between "something I can fix myself" and "something only a human should decide." So it did neither — it just let those things pile up quietly. I built it an escalation signal: now, when the autonomous side hits something structural, it raises a flag that the next session sees immediately. Honest iteration, again. You find where the judgment falls short, you fix it, you let it run.
The bigger shift is in how I think about what comes next. I'd been treating DARA's roadmap as a list of features — memory, then eyes, then ears. That's the wrong frame. The right frame is a body with senses.
There will be one common channel — a facts layer — where everything DARA perceives lands as small, atomic facts. The autonomous side curates that channel: it decides what's new, what's already known, and what contradicts something it already holds. And then each sense is just an input into the same channel. Reading is one. A memory connection for other AI tools to contribute through — via something like MCP — is another. Screen and audio come later. Same channel, one curator. This is context engineering at its most literal: designing what context reaches the model and how.
A week ago DARA learned to keep itself tidy — 140 times over, without me. Next it learns to perceive.
— Javier
EIDARA v2 is free. SUPER DARA is what comes next.