Both houses open their energy screen on light. Nothing unusual there.
But when we lined up 28 days side by side, one house retracted the screen later in the morning and kept a small residual as the sun came up. Less venting into the rising light → a faster morning humidity drop → higher, more frequent midday VPD peaks. On the two most extreme weather days the houses were identical, the gap only showed on the moderately hot days, which is exactly where climate strategy usually lives.

Easy conclusion: "House A is steered more generatively." Maybe. But here's the catch worth pausing on:
Each screen opens on its own outdoor radiation sensor, up on the roof or the weather mast, fully exposed to the elements.
Out there a sensor lives a hard life: dust, pollen, a film of grime, rain spotting, the odd bird. Add a few degrees of tilt, a slightly different position on the roof, or a structure throwing an early-morning shadow across one and not the other, any of these makes it read low. The controller then "thinks" it's darker than it is, and holds the screen a little longer. One outdoor sensor, quietly a few percent off, can shift a whole morning of climate inside and it shows up in the crop as VPD stress, not as an alarm.
So before we credited (or blamed) the settings, the honest first move was: are the two sensors even seeing the same light? Same mounting position, same orientation, same cleaning schedule, same calibration date, two exposed rooftop sensors are exactly the kind of thing that drifts apart over a season. Until that's confirmed, part of a house-to-house difference could simply be two sensors telling two different stories about the same sky.
The wider point: automation doesn't fix a bad measurement, it amplifies it, faithfully, all day, every day. A sensor's data is only ever worth as much as its installation and its cleaning. The cheapest climate improvement on many sites isn't a new strategy. It's a clean sensor and a ladder.
How often do your light and climate sensors actually get cleaned and cross-checked against each other?