The Story · Build in Public

We built the wrong product first.
Here's the honest version.

8 months. One beautiful analysis tool nobody needed. One overnight commoditization by big AI. And the pivot that turned it into something artists actually use: a song in, finished content out.

Late 2025

Eight months ago: we start building the pipeline

A song-analysis engine, stage by stage — tempo, key, structure, hooks, lyrics, instruments. Test material: our own songs. If the models couldn't explain why our best track was our best track, they weren't done.

Winter 2025/26

Months of collecting a legal song-outcome dataset

You can't validate a Virality Score with vibes. So we built a fully legal dataset of song metadata and real streaming outcomes — hundreds of thousands of tracks over months — to check every scoring idea against what actually happened in the market.

Spring 2026

Going online

The website ships. An early-access waitlist opens. First artists sign up before the product even exists — the promise alone (know what's viral about your song before you release it) was enough to get people in the door.

May 2026

The app goes live

Upload → full analysis in about a minute. Virality Score, best moments, song DNA, lyric scoring. We tested relentlessly, shipped fixes daily, and onboarded our early-access artists.

June 2026

The honest part: it failed

Barely any user feedback. The funnel wasn't good enough — people looked at a beautiful analysis, said "cool", and left. And the market moved under us: general-purpose AI like Google's Gemini made "tell me about my song" a free chatbot prompt. Pure analysis was commoditized overnight. A score alone doesn't post itself.

July 2026

The pivot: analysis → content, one click

We stopped selling the analysis and started selling what it enables. Now one upload returns 7 ready-to-post reels cut to your strongest seconds — beat-synced, word-accurate lyrics — plus captions and a 2-week posting plan. The analysis is still there. It's just the engine now, not the product.

Today

Specializing in what content works for music

Every posted reel you track teaches the system which angles work for which sound. That's the road ahead: analyze a song and know — from real outcomes — what kind of video works for music like yours. Good content in seconds, for artists without a budget, no prompting skills required.

Why we exist now

Most musicians can't afford an editor, don't want to learn GenAI video tools, and shouldn't need prompting skills to promote a song. The tools exist — they just cost €100/month and a weekend of tutorials per release.

Songbrain's job is simple: good content in seconds, from nothing but your song.The analysis pipeline we spent 8 months building didn't die — it became the engine that knows exactly where your song is strongest and cuts the videos there. And every reel you post and track teaches it what kind of content works for music like yours.

Chapter next: your song.

Free to try, no account needed. Upload a track and watch the reels render while the score comes in.

Analyze your song free →