What breaks first when you turn a long interview into show notes?

I’m working on a small browser-based workflow for turning raw interviews and podcast episodes into a timestamped transcript, a short summary, show notes and subtitle drafts. The hard part does not seem to be generating text; it is keeping names, numbers, timestamps and speaker changes straight across a long recording.

For people who edit interviews or podcasts: what breaks first in your current workflow when the recording is 30–90 minutes? Names, numbers, timestamps, speaker changes, noisy audio, or turning the transcript into useful show notes?

If you have a public episode clip you are comfortable sharing, I’m happy to turn one 30–60 second section into a transcript + SRT draft and report what still needs human review. No private upload needed.