He was likely using the ASI224 as a guide camera.
In astrophotography, you tend to get a lot of noise build-up in images. You can reduce the noise by collecting more samples of data. The noise can be reduced based on the square root of the number of samples. e.g. with four shots of the same thing, you can cut the noise down to by a factor of 2 (or 1/2 of the noise). Take 16 shots and you can cut it down to 1/4. The problem with this is that you start to hit diminishing returns (hours and hours of more data starts to make only a very tiny improvement)
There's another issue ... which is that a lot of frames may need to be rejected due to vibrations, atmospheric distortions, etc. Stacking software such as AutoStakkert can help you analyze the frames and select say... the best 10% of them.
So basically you're collect a TON of data, knowing that most of it will be rejected ... and then of the data that isn't rejected you want enough of it so that the noise reduction.
The sunlit area of the moon is a fast exposure. So fast that you wont see any stars at all. Also the unlit area of the moon will be completely black. To get the stars to show up, you need a long exposure. To get the unlit side of the moon to show up (via "Eartshine") you need a longer exposure. But there are problems with tracking accuracy of the mount when doing this. SO... you use a 2nd camera as an "auto-guider".
When tracking a guide-star, the guide camera has a reference frame and then each successive frame is compared to that reference frame (it is typically taking an image every 2-5 seconds). If the guide-star moves by even a sub-pixel amount, the software will send a command to the mount to nudge it back on track.
The moon doesn't cross the sky at the same rate as the Stars since the Moon is in orbit around the Earth. But there is guide software that can guide on the moon.