appscript.dev
Blog Drive Docs

How I automated my podcast publishing checklist

From Northwind's recording to release — chapters, show notes, transcript, in one chain.

By Awadesh Madhogaria · Published Sep 26, 2025

Northwind runs a small podcast, and recording the episode was never the problem. Everything after the recording was — the transcript, the show notes, the metadata, the dozen small steps before release.

Publishing an episode meant a long, dull sequence of small jobs: upload the audio for transcription and wait, write show notes from scratch every time, then generate a cover image and fill in the episode metadata. None of it was creative work. It was the kind of repetitive checklist where steps get skipped and quality drifts — some episodes had great show notes, others had two rushed lines.

That post-production tail quietly ate most of an episode’s effort. So I mapped the whole checklist, asked which steps were genuinely mechanical, found that almost all of them were, and turned the list into a chain that mostly runs itself.

How the chain works

Dropping a file into a folder kicks off the whole thing — by the time I look again, there’s a draft episode waiting:

  1. An audio file is dropped into episodes/incoming/.
  2. The script uploads it to my transcription provider.
  3. The transcript is saved to Drive.
  4. Claude generates show notes from the transcript.
  5. A cover image is generated by a separate flow.
  6. Episode metadata is written to a Podcast sheet.
  7. The one manual step: I hit publish.

The script does the mechanical chain and hands me a finished episode to approve. It picks up the audio automatically, drafts consistent show notes, and assembles the metadata so nothing is missing at release time. The one thing it doesn’t do is decide.

Where I keep a human

The “Should this episode be published?” decision stays with me, and that boundary is deliberate. A script can assemble an episode; it can’t judge whether it’s any good. Transcription quality varies too, so I still skim the transcript before show notes go out. Keeping a human at step 7 means a bad episode never ships on autopilot.

Was it worth it

Yes — maybe 90 minutes saved per episode. But the consistency is the part I didn’t expect to value so much:

  • Every episode now has proper show notes, not a rushed two-liner.
  • The metadata is always complete, so nothing breaks at release.
  • The boring tail of the job no longer makes me dread publishing day.

Automate the assembly, keep the judgement. A publishing checklist is mostly mechanical steps with one real decision at the end — let the script do the chain and save your attention for the call only you can make.