I've dabbled a few times with descript. Here are my experiences and what I think of it so far:
I've run episodes through descript to get a transcript of them many times. It's very helpful for clients if they want a transcript for their podcast. Descript also has an option to render subtitles with a video, which you can easily overlap onto your video.
Descript doesn't take too long to process a transcript. It took about 5 minutes to process a 45 minute long podcast interview.
In descript you can go through and "scrub" out a podcast. Scrubbing means removing all of the filler words, breaths, repetition, etc. Pretty much the long time manual work a podcast sometimes needs.
Descript is famous for it's auto filler word removal. With the help of the transcript, it can detect almost every filler word in an audio file, including ums, uhs, and you knows. Descript will underline any detected filler words for you when you look through the transcript. By right clicking one of the filler words, there's an option to auto remove them all. This process can take longer than the transcript to process. I recently processed a whole podcast episode through its auto filler word removal... I would strongly advise against doing this without a double listen to make sure it sounds good. It constantly messed up edits and many sentences that had a filler word would sound really choppy. It wasn't too good at detecting more complicated mess ups in the call, like if a guest or host repeated a sentence (which I would usually take out).
I've learned the way to go is process the transcript, then rather than auto removing the filler words, go to them (descript provides a list of the filler words with time stamps) and manually remove them.