How Claude Produces My Hospital Radio Show
I’ve been presenting my hospital radio show for a while now, and I love every minute of it.
Choosing the songs, setting the mood, thinking about the patients listening from their beds (and the staff listen too).
It’s genuinely one of the most rewarding things I do. But there has been one part of the process I quietly dread: building the playlist.
🎬 The Problem with Prep
Before every show, I sit down to work through what I’ve played recently. Unlike some commercial radio that won’t repeat songs within an hour, I want to make sure I’m not repeating songs within 4 weeks!
I had already built a live dashboard (built with Claude), which help me to search by song and show
Had I played that track from Les Misérables two weeks ago or four? Was Unchained Melody from Ghost going to feel fresh, or had I overplayed it?
I’d scroll back through old playlists, cross-reference, second-guess myself, and generally spend at least an hour, sometimes more, just trying to avoid repetition.
It wasn’t the creative part. It was admin. And frankly, admin is nobody’s favourite job.
📊 Expanding the Role of Claude
The Google Sheet I use to log every show is a decent record, but I wasn’t really using it. I was just filing it away and then doing the mental gymnastics every time a new show came around.
That changed when I built a Claude AI skill that actually reads the spreadsheet, thinks about it, and does the heavy lifting for me.
Here’s what it does:
- Reads the full song history from my Google Sheet (each tab represents one show)
- Analyses which tracks have been played and when
- Ignores anything from the last four weeks — so recent songs are automatically off the table
- Builds a fresh playlist suggestion for the next show, pulling from tracks that haven’t been played in a while
It’s not random. The four-week window means the show always feels varied without me having to keep a mental tally of every track I’ve ever played.
⏱️ The Time Saving Is Real
What used to take an hour or more now takes a few minutes. I open Claude, run the skill, and I have a playlist ready to review.
I still check it over, tweak a song here or there, and put my own stamp on the running order. But the initial groundwork is done.
That’s the thing about AI that I think gets missed. It’s so much more than a quick chat.
It’s also not about replacing your judgement.
It’s about removing the tedious stuff before you get to use your judgement.
The decisions are still mine. I’m still the one who knows that a show needs to open with something uplifting, or that a slower ballad works better mid-show than at the end.
AI just hands me a clear starting point instead of a blank page and an hour of scrolling.
🎵 What It Means for the Show
For the patients and staff (and anyone listening online) of course, none of this is visible; and that’s the point.
They just, hopefully, hear a well-paced, varied show with songs they might not have heard for a while. No accidental repeats. No presenter quietly panicking that they’ve played Circle of Life three times in six weeks!
For me, it means I can spend that recovered hour on the bits that actually matter: writing better links, researching an interesting fact about a song, or simply going into the show feeling prepared rather than frazzled!
🤖 The Bigger Picture
I’ve written before about how AI fits into my day-to-day life — from managing my Cub Scout pack to drafting articles like this one. But the hospital radio playlist tool might be the most quietly satisfying example yet.
It solves a genuinely annoying problem in an elegant way.
If you’re doing anything creative on a regular schedule, writing a podcast or a newsletter or prepping for a radio show, and you’re keeping records of what you’ve done, there’s almost certainly a version of this that would work for you.
The data you’ve been logging? It’s more useful than you think.
You just need to give it to the right assistant.