Half of these AI apps will fail

Happy Christmas Eve. Hope everyone has a great time this holiday season with families or friends.

Today we are talking about AI apps. Which ones? All of them. This feels like Crypto 2021 all over again.

So this all started with a tweet I saw from A16z partner, Justine Moore. I have been following her work since college and highly recommending following her. Nonetheless, here is the tweet:

How the hell are all these apps and companies going to “win”? Well, they won’t.

Let me start by saying that I am a big proponent of AI. I used ChatGPT to build an entire website using python and I have never coded a day in my life. I believe the technology is revolutionary and is just starting to be used properly. The technology isn’t going to replace jobs, but rather make us more efficient in whatever we do. That website I built? 506 lines of code completed in under 30 minutes.

So here’s the problem:

There are plenty of tasks and daily workflows that can be improved with AI. They can become more efficient, be done better, and allow for humans to do the more important tasks. Stuff like note-taking during meetings, editing videos, make creative ads, or doing cold outreach. All of these tasks are now being replaced with AI apps. The problem is that there are maybe 4 or so AI notetakers for meetings, 5 video AI editing/generating softwares, and 3 AI avatar apps. This is just on the surface level, but point is these apps are not venture back-able businesses. You cannot bet that one AI notetaker app is going to win the market and produce venture-scale returns.

If you are a founder who bootstraps these types of apps, this can be a goldmine. You can build AI apps nowadays that produce hundreds of thousands in monthly recurring revenue with little capital. I have seen studios who have created 7-8 AIs and have millions and annual revenue. These are just not businesses who can do hundreds of millions in revenue a year.

How can they?

It depends. I think you can take a measured approach or get lucky, but you can’t start by doing 5 things well. You need to start by doing one thing really, really well. Become trusted by a lot of people in one space and build up a cash cow of a product. From there you can grow into similar segments, or in other words, go vertical.

Verticalization

If you know a VC who works in SaaS you have probably heard this term before. Vertical SaaS is idea of adding new solutions and products in your companies that relate to your specific industry or problem you are solving. Let’s take an AI notetaker for example. If they wanted to go vertical, maybe they would create an AI communication tool next.

Most of these apps are trying to be really good at one thing, which is the right move. Eventually though, they will need to expand and build more products. After all one of the most important metrics in any SaaS company is Net Retention Revenue (a way to measure how good you are at adding new products to your current customer base).

The issue I foresee with most of these apps is they are going to be trying to create new products in spaces where there are already 4 products. You are going to have multiple companies fighting over verticals all throughout the market.

Final thought

So to state to obvious of the venture model, yes most of these companies will fail. Most others will be bought by incumbents, and a very small select few will become billion dollar businesses. I would encourage founders and entrepreneurs to not raise for as long as you can. Bootstrap and build your cash cow product, then fundraise and have all the leverage.

Happy Holidays.