Capturing screenshots and recording audio in an ongoing basis, and submitting data to a RESTFul API

Hi there - know this might be a silly question, but asking anyway... As in, know these formats/data-types are probably not really possible to compress any more than they already are.
Have managed to sort out capturing screenshots repeatedly, while recording audio in the background, using combination of PIL's ImageGrab, and pyaudio, and can then use moviepy, which is a sort of wrapper around/interface to the FFMPEG command-line utility - this all allows me to record forms of screencast recordings, setting my own forms of time-frames, etc. in terms of the looping interval when I want to capture screenshots, etc., before then combining them into video clips with the audio recording merged in as a background track, and, all works fine, but, we want to use this as a form of monitoring service for call-centre staff, at times, and, the only real remaining issue is file-size/data in terms of both hard-drive storage space, and, bandwidth in terms of submitting resulting data to a RESTFul API.
For example, a test video clip, generated using the libvpx codec, resulting in a .webm file, with a total length of 14 seconds, has a file size of 100KB.
Also, don't think it's really relevant, but, am then just using things like requests module to submit data to the RESTFul API, which have used flask to implement.
So, know this question might be a waste of time since have already played around with selecting the video codec that generates the smallest resulting file-size, and, not sure if might be able to drop image snapshot file sizes by using something like grayscale, which moviepy doesn't want to work with directly during generating original video clips, but just wondering if there might be any way to try converting binary data into smaller data chunks to then upload these via my RESTFul API, where could then convert them back to multimedia formats, etc.?
Any thoughts/suggestions on this type of thing, and, on that note, all of this will be running as something like a background service on call-centre staff's windows 11 machines, if relevant.  As in, if there might be some way to store data and then generate multimedia later on, on the server handling the RESTFul API, that could also work, but, main thing is to both save storage data on workstations, as well as limit amount of bandwidth required overall since the number of target machines could easily be enough to use up a lot of bandwidth, etc., so, what we are looking into at the moment relates to only triggerring recordings at certain times on certain machines, in between.
Thanks in advance
---
Jacob Kruger
+2782 413 4791
"Resistance is futile!...Acceptance is versatile..."