As I search for a cofounder for Remake.ai, I've been talking to founders of AI startups that recently ran out of runway.
Surprisingly, many of these startups had traction. In some cases, they had a product–market fit.
What went wrong?
A lot of these companies were AI wrappers — web apps wrapped around online LLMs like ChatGPT or Anthropic.
Here are a few AI wrapper type startup ideas:
"AI patent attorney"
"AI financial advisor"
"AI recruiter that ranks job applicants"
Need more ideas? Prompt ChatGPT "Generate ideas for AI startups."
The problem with AI wrapper startups is - while they're easy to build, they often have no competitive moat.
Let me give you a worked example. Suppose I wanted to build PatentAnything.ai, an AI startup that helps you patent anything, almost instantly.
PatentAnything.ai takes your idea, checks if it's patentable, and—if it is—writes the draft, adds drawings, handholds you through the USPTO filing process and collects a convenience fee.
"Save big on patent attorney fees! File your patent in minutes, before someone else does!", PatentAnything.ai pitches to the (apparently experienced) inventor.
If your idea is NOT patentable, AI suggests creative ways to MAKE it patentable. "Make your idea patentable using AI instantly!" PatentAnything.ai pitches to the (apparently amateur wannabe) inventor excited to and, hopefully, willing to pay the convenience fee at the spur of the moment.
Here's how I would build PatentAnything.ai:
Frontend in Next.js
Backend calling OpenAI or Anthropic APIs
Stripe API for payments
Hosted on Vercel with serverless functions
All code generated by Claude Code from a single prompt
That's a one-day MVP. But that's also the point: if I can build it in a day, so can someone else.
Disclaimer - PatentAnything.ai is a trivialized example. You can build a competitive moat using other methods. For example, you can clone Facebook technically but never outcompete it due to Facebook's network effect moat.
Lack of competitive moat makes businesses fail. Importantly, one can build an AI wrapper, find traction, raise capital and appear to be a successful startup - only to find that your success has attracted copycats who compete with you on price, changing your race to the top to a race to the bottom.
This is why I'm building Remake.ai, NOT an AI wrapper startup.
You may ask, why makes Remake.ai so special? How is it different from AI wrappers? What is Remake.ai competitive moat?
Answering this question may require a separate post. I believe Remake.ai can be for consumer robots what Apple is for smartphones.
Interested? Let's talk more! Drop me a message.
Subscribe on Substack to get updates when new posts are published.