How to Survive the Post-AI World (As An Online Business)

I started neural frames in December 2022. Whenever people asked me about the use case for these AI animations, I’d just shrug and say, “I can’t be the only person who enjoys making these - there must be others, too.” In other words, I was coming entirely from a technology point of view: here’s a cool technology to create something, so I’ll build a product to use it.

If you read up on how to come up with startup ideas, you’ll learn this is a very common pitfall. Now that I’ve been in the startup world for a while, I see it constantly when talking to folks who want to start a company.

Because here’s the problem: unless your customers themselves are extremely technical, they don’t care about your technology. They really don’t.

Initially, neural frames was built for everyone interested in the tech, with the tagline: “Text-to-video for everybody.” I got some early traction, and I’ve written elsewhere about talking to those first users - who ranged from people wanting to visualize their dreams, to children’s story authors, app developers, short-story writers, and musicians.

But after a few months, growth stalled, and I wasn’t sure how to move forward. It took a long afternoon stroll in Medellín (why did the most insightful moments of my life always happen during walks?) to figure out what the platform should really look like and who it should be built for. That’s when I realized there was massive potential in niching down to the use case I know best: musicians needing music videos.

I think this was a pivotal moment in our history, and here’s why:

On paper, it sounds great to be super broad with your target audience. The more people you build for, the more potential buyers, right? Right???

That’s a trap - and it’s only getting bigger in the AI era.

If you niche down, you discover wonderful things. Marketing to a specific group is way easier than marketing to everyone everywhere. Tony Fadell wrote in Build that they’d defined their ideal customer persona down to a name and hair color. With every product decision, they’d ask (paraphrasing), “Will Jack from Minnesota like it?”

It’s true. As soon as I focused neural frames on musicians, life got so much easier. I could build features the dream-visualization crowd might hate but musicians would love. I could change the tagline to “The synthesizer for the visual world,” and I could start ranking for music-focused keywords.

Then ChatGPT released its image model last week, and I got a little freaked out. Even though music videos are pretty niche, they’re visually appealing and widely recognized. I’d bet OpenAI will launch a feature within the next 1.5 years letting you upload an MP3 to ChatGPT and ask it to make a music video in a certain style.

So I wondered, “Will this be the end of neural frames?” But I don’t think so.

We’re niche. We know exactly who we’re targeting, and we’re building a platform full of features they want. Like lyrics-to-prompt conversion, beat detection, audio-reactive effects, etc. ChatGPT won’t go down that specific route.

Here’s my prediction for at least the next 10 years:

While we’ll soon live in a world of hyper-smart AI assistants that can do basically anything, they won’t be able to outdo tools that make their living from solving one specific task for one specific niche.

AI Music Video of the week

One of our users wrote in the nicest words about neural frames and how we’ve “changed their art completely”. After many videos with the realistic looking Kling and Runway models, it was very refreshing to a see frame-by-frame video made for original (classical!!) music again.

What do you want to read about next? Hit reply and let me know! 💜 

Have a great week.