I Replaced My Spotify Discover Weekly with Songs I Made Myself

Every Monday, Spotify gives me 30 songs it thinks I'll like. I listen to maybe five. I save maybe one. Most weeks, zero.

Last month I tried something different. Instead of waiting for an algorithm to understand my taste, I spent 30 minutes making five songs in the exact style I wanted to hear. Warm acoustic singer-songwriter with my actual life woven into the lyrics.

I haven't opened Discover Weekly since.

The Realization

I was listening to a Jake Scott song - one of my favorites - and thought: I wish there were more songs that sound exactly like this but about my life specifically. About my daughter learning to ride a bike. About the drive from Liberty Hill into Austin. About the weird satisfaction of watching your kid finally get long division.

Jake Scott will never write those songs. No artist will. Those are my moments. The only person who can turn them into music is me.

So I did.

The Process

I'd already built an AI artist profile - a singer-songwriter named Cormac Riley, designed to blend the sounds of Jake Scott, Ben Rector, and Matt Kearney. He has three Suno prompt templates for different song types: upbeat, ballad, and mid-tempo.

Monday evening, I sat down with Claude and brainstormed five song concepts based on my week:

Song 1: Noelle reading with a flashlight past bedtime. Warm and amused.

Song 2: The feeling of driving through Hill Country at golden hour. Peaceful and grateful.

Song 3: Charlie asking why the moon follows our car. Sweet and funny.

Song 4: The stress of building something new and not knowing if it'll work. Honest and hopeful.

Song 5: A slow song about watching my wife Ashley laugh at something across the room.

For each one, I wrote lyrics using my lyric-writing framework - consistent syllables, rhyme schemes, names at the start of lines, open vowel sounds on chorus endings. Each set of lyrics took about 5 minutes with Claude's help.

Then I generated. Two or three versions per song. Picked the best. Total time: maybe 30 minutes.

What I Got

Five songs about my actual life. My kids' names in the verses. My town in the imagery. My emotions in the choruses.

And they sound good. Not "good for AI" - good. The vocal is warm and consistent. The production is clean. The melodies are catchy. My wife heard the one about her and got emotional.

These aren't songs I'd skip after 10 seconds. They're songs I genuinely enjoy listening to. Because they're mine.

Why This Hits Different

There's a music psychology concept called the "mere ownership effect." Things you create or own feel more valuable to you than objectively identical things you don't own. It's why your kid's crayon drawing means more to you than a museum print.

Custom songs amplify this by an order of magnitude. A song with your daughter's name in it, about a moment you actually shared, in the genre you already love? Your brain can't not engage with it. It's personal in a way no algorithm-recommended track can be.

I play my five songs more than I play anything Spotify recommends me. That's not because they're objectively better than what professional artists produce. It's because they're objectively more relevant to my life.

The Unexpected Benefit

My kids now associate songwriting with something we do together. Charlie suggests topics. Noelle critiques lyrics - "Dad, that line doesn't rhyme." They hear themselves in the songs and light up.

We're building a catalog of family music. Songs about our actual experiences, our inside jokes, our milestones. A soundtrack to our life that didn't exist before and wouldn't exist without AI.

Ten years from now, when Noelle is in college, she's going to hear a song about the night she read Harry Potter with a flashlight under her covers - and she's going to remember being nine. That's worth more than any Discover Weekly playlist.

You Don't Need to Go This Far

The full artist-building system I use takes a couple hours to set up. But you can make a single great song in 30 minutes.

Pick a moment from your week. Write four verses about it - keep lines the same length, make them rhyme. Add a chorus that captures the feeling. Format with section tags. Use a Suno prompt with specific genre, vocal, and instrument descriptors.

Generate. Listen. Iterate if needed.

You'll have a song about your actual life that sounds like it belongs on Spotify. And you'll understand why I don't bother with Discover Weekly anymore.

The artist-building system, lyric framework, and prompt libraries I use to make these songs are the core of a course I'm putting together. If making your own personal music sounds like something you want to learn, the course is live now at ideatomusic.com/course.