Steal My Idea #18: An App That Puts You Inside a Music Video

Hey buddy,

Picture this.

You upload two photos: you and your partner.

You pick a music video aesthetic. Cinematic sunset field. Rainy rooftop dance. Slow-motion beach walk.

And 30 seconds later, an AI-generated video comes back where the two of you are the ones in it. Same lighting. Same cinematography. Same emotion. Just your faces.

You download it. You open Instagram Reels. You add Perfect by Ed Sheeran. You post it.

Your friends lose their minds.

That's the app. And nobody has built it properly yet.

I'm not a mobile developer. I can't build this.

But I've thought about what makes something go viral on Instagram and TikTok for a long time.

And this hits every single trigger.

It's visual. It's personal. It's shareable.

It makes the person who posts it look creative without them having to be creative.

And it has an obvious emotional hook: love, relationships, milestones - that makes people want to share it the moment it's done.

The best consumer apps make you feel something in the first 30 seconds. This one does.

The product is a mobile app: iOS and Android, that generates a 30-second AI music video starring whoever uploads their photos.

Here's how the experience works:

Step 1 — Upload your photos. The user uploads 3-5 photos of themselves and their partner. The more angles and lighting variations, the better the AI can maintain face consistency across the generated video. Clear face shots work best and the app tells them this upfront.

Step 2 — Pick a scene aesthetic. Not a song, a visual style. This is intentional. The app offers a curated library of cinematic scene aesthetics: golden hour field, city lights rooftop, slow-motion rain, cosy café, vintage film, dramatic desert, underwater dream. Each aesthetic has its own lighting, colour grading, movement style, and emotional tone. Users pick the one that matches the vibe they want.

Step 3 — Generate. The AI renders a 30-second video. The couple appears in the scene, dancing in the rain, moving, looking at each other, living inside the aesthetic they chose. Face consistency is the entire technical challenge here and the entire product promise. Distorted faces kill it. The 30-second format exists specifically because it's the sweet spot where current AI video generation can maintain coherent, recognisable faces throughout. Longer videos break the illusion. 30 seconds nails it.

Step 4 — Download and add your music. The video exports clean, no audio. The user opens Instagram Reels or TikTok, adds Perfect by Ed Sheeran or whatever track fits the vibe, and posts. Instagram and TikTok have already licensed most popular music. The app never touches the audio. This app actually tells you the ideal timestamps of the music to be inserted for the perfect flow of the video.

This isn't a workaround. It's a feature. "Make the video here, add the music on Instagram" turns the distribution step into the natural final act of the experience. The user posts directly where their audience already lives, with the music already licensed. Clean, legal, frictionless.

Couples are the core. Anniversaries. Valentine's Day. Proposals. The "I just wanted to do something different for our 2-year" post.

The emotional occasions where people want something that feels cinematic and personal but don't have a film crew.

This is the product for that moment.

Singles and creators are the secondary use case. Solo dramatic Instagram content. A friend group doing something funny or creative. Anyone who wants a 30-second AI reel that looks genuinely impressive without filming anything. You can star in a kanye west styled video and look cool af.

One-time generation at $4.99 per video — low enough that it's an impulse buy, high enough to cover compute costs. Couples on an anniversary aren't price-sensitive for something this personal.

Subscription at $19/month — unlimited generations, priority rendering, early access to new aesthetics. For creators who want to make these regularly.

The one-time model is the entry point. Most users buy once for a specific occasion. A percentage convert to subscription when they realise they want to keep making these.

The viral loop: Every video posted on Instagram or TikTok with a subtle "Made with [App Name]" tag is a free ad. One viral post from a couple with 10,000 followers brings in 50 new users. The product markets itself if the output is good enough. You don't need an ad budget, you just need the first video to be genuinely impressive.

AI video generation has crossed a threshold in the last 12 months.

Tools like Runway, Kling, and Hailuo can now generate short-form video with consistent characters across scenes, something that was impossible two years ago.

The 30-second format is specifically chosen because it's where the technology is reliable today. Face consistency across 30 seconds is achievable. Face consistency across 3 minutes is not yet.

The window is this: the technology just became good enough, and nobody has built the consumer-facing product around it yet.

There are AI video tools for professionals and creators. There is no dead-simple "put yourself in a music video" app for regular people.

That gap closes in 12-18 months as the big platforms catch up. The time to build is now.

The first video the user generates has to genuinely impress them.

Not "pretty good for AI." Actually impressive. The kind of thing where they watch it back and immediately want to show someone.

If the face looks even slightly off: warped jaw, inconsistent eyes, wrong skin tone, they close the app and never come back.

If it looks real, they share it immediately and tell three people about it.

Every technical decision in this product: the photo upload guidance, the 30-second limit, the scene curation, the render quality exists to protect that one moment.

Build the quality bar before you build the features.

Start by building on top of Runway's API or Kling's API, both are available to developers today. You don't need to build the AI from scratch.

Start with one aesthetic. Perfect it. Make it genuinely impressive. Launch with that one scene and expand from there.

The app doesn't need 50 aesthetics on day one. It needs one that makes people say "wait, how did they do that."

Go.

If you do not know coding but still want to explore this, Here’s a no-code AI app builder with an exclusive offer for Wifi Moolah readers:

Talk soon, Kris

P.S- The best consumer apps don't solve problems. They create moments. A couple watching their AI music video for the first time is a moment. Build for that.

If you’ve read this far, connect with us: