Eudaemon

To what extent are informed decisions better than instinctual ones? Eudaemon turns a person's records into evidence strong enough to act on, and turns the goals that person brings into questions that evidence can answer.

We all leave a trail of data behind us; smartwatches, banks, inboxes, medical files. If you claimed every record you can lawfully obtain and fed them to a machine, could it learn what you actually do? Could it take your goals and show what you'd have to change to achieve them?

Eudaemon is one person finding out. Every record is claimed and lands in an encrypted archive (the data lakes). Over the lakes sits the evidence layer (the deep twin), which labels each record and makes it findable. Claims about the individual are hypothesised, corroborated and given a confidence rating. A goal layer (Telos) splits goals into questions the record can check. An experiment engine exists to prove causation from any correlations the twin uncovers, by running small randomised trials. The agent layer (Jarvis) isn't built yet, but is envisioned to automate the schlep work required to achieve goals, in a manner personalised by the twin. The latest progress on the project is in the log, most recently entry 02.

Now: Entry 02 is live. The assumptions are now ranked tests, and the first measures how strongly one day predicts the next, source by source; its answer decides whether Eudaemon reads records or runs experiments.  ·  updated 30.08.2026

The log