Having spotted the numbers/letters on the pins weren’t what I was seeing in my head, here’s a small simulation that actually confirms the original assumption, the LiPo will slowly discharge through the ADC divider until it reaches the difference between the 3.3 and the exact value that that half of the AO7801 is at (ie real world Vgs between -0.5 to -0.9V)
You can see the step at 3.9V, at 4.1V the drain is 19uA
The thinking behind this will be that LiPo’s tend to settle quite quickly to ~3.7V so it’s not losing too much energy, but still a bit of a waste.
I personally use BSS84’s after a disappointing incident with KiCAD. They have a Vgs of -1.6V so I’ve not really thought about the potential (pun intended) for loss of potential.