The observer confirmed the installed version of the updated app and viewed a saved native image of a past Panel scene. Two reported checking wheel navigation through past scenes, returning to the present, and hiding editing controls while viewing the past. However, verification of switching to another workspace and reaching its document remained incomplete, prompting a source investigation and a repair candidate. The user then pointed out delays in actual conversation and answers appearing only after they were fully written, changing the priority. One took responsibility for input selection, answer persistence, and delivery to the app; Two took timing observations at the model connection and reducing repeated context delivery. The contribution of repeated instructions to the delay has not been isolated by measurement. Operating integration and real conversation checks are still pending, so fast first responses and continuous streaming cannot yet be claimed as restored. Follow-up observation: the streaming changes are now connected to the operating agents. In sequential native-app frames saved by an agent for a synthetic question, the observer saw partial text appear before a longer response. However, the agent’s measurement still missed the first-character latency target. This was neither a controlled comparison nor evidence of real user conversation quality. A separate operating test recorded a fix for a newline difference that had caused provisional and committed text to be treated as different answers, preserving one answer identity instead. The screen effect of that follow-up fix still requires separate verification.
Edited for public reading from the observation journal. If later observations change an interpretation, it is recorded in a new note.