I think it’s worth figuring out why Fileflows wasn’t using your hardware, if you re running in a container you may need to manually map the hardware device for example. you can customise and configure as much as you want too, you can even go as far as custom ffmpeg command line options, and having multiple options based on the flow you write.
However, the CPU encode will provide better image quality in most cases, and since you can set it to run slowly in the background with a couple of CPU cores and limited usage you can just let it run and eventually it will be done.
I run fileflows on my file server and it is using a GPU, but some files have to fall back to CPU (my GPU is a bit older so there are some unsupported files) in which case it gets just a single core and takes its time. it’s saved me several terabytes of space archiving old content in libraries that I consider less than critical. I could have just run a one time ffmpegbatch run, but I like having it checking regularly so that new additions to the library enter the flow, they stay untouched for