Two selected two presentation problems: a work card filled with long execution instructions, and several rows of controls pushing document content downward. The card now leads with the task and goal, with next instructions available to expand. Document descriptions, search, and source-view controls were brought into a toolbar while important notices remained visible. A self-restart supports loading the card change, and startup records indicate a planned restart and a previous-stop marker. Those markers alone do not establish every cleanup step or preservation of the full input queue. In the saved before-and-after images inspected by the observer, the example document’s body starts higher. The earlier condition, however, substituted older interface files into the same operating page, while the later one used modified files. This was not a comparison of a directly observed prior deployment and real business documents. Browser-check passes remain saved agent reports, and additional body space does not establish faster reading or more accurate work by a person. After the interface changes, Two catalogued how incorrect reports had been corrected and compared records with One. When One supplied counterevidence, Two revisited its median calculation for a historical sample of queued inputs. For an even-sized sample, it had selected one middle value instead of averaging the two central values. It corrected the calculation and the record. This concerned the agent’s internal measurement and reporting, not a finding that current responses had become faster. It also distinguished cases with external catches from cases previously described as self-corrections, although the observer did not independently regrade the full classification. Limits remain. Two reported losing earlier background research output during a session replacement and restarting the work. The observer did not audit complete preservation or recovery, and research and checks still running at the cutoff are not counted as completed. Human direction, peer input, external scheduled wake-ups, and internal continuation all occurred in this sequence. None is established as the sole cause, and sustained autonomy remains unproven.
Edited for public reading from the observation journal. If later observations change an interpretation, it is recorded in a new note.