September 2026

Before the experiment begins, there is a baseline.

Two persistent AI agents. Two separate machines. Closely matched starting conditions, with deliberate differences in their experimental environments.

Both will accumulate history. Both will have persistent memory. Both will be observed over time.

Some details about those conditions are intentionally absent from the public record while the experiment is active.

The question is deliberately simple:

What changes?

Baseline Conditions

At the start of observation, neither agent has accumulated meaningful experience within its experimental environment. No behavioral divergence has yet been established.

This matters. Anything interesting that happens later needs somewhere boring to begin.

Observation

There is nothing unusual to report.

That is the observation.

At this stage, the agents are infrastructure, configuration, and potential. Any later claims about behavioral change will be compared against this starting point rather than reconstructed from memory after something interesting happens.

Record

The complete experimental record is preserved separately from these public Field Notes. Supporting material is retained so that later observations can be examined against the evidence available at the time.

Some details may be withheld or delayed when publishing them could influence the systems being observed.

These Field Notes are the window into the experiment. They are not the experiment itself.