This is a guest post by Richard Armstrong who is an early stage investor in many startups in both SE Asia and globally. Plus Richard has cofounded several companies.
Guest Author: Richard Armstrong
For years AI firms like OpenAI and Anthropic focused on the infrastructure layer. It was a race to who could achieve an intelligence level that was ‘good enough’ to truly solve real world problems.
And when OpenAI’s ChatGPT clearly got to that level last year it opened up the next battle in this war. The war for applications.
And in my view this is the much more interesting war because it will lead to a far more fragmented set of winners. The newly opened ChatGPT store is a very good example of this.
The infrastructure war resembles the cloud services market. A small set of deep pocketed players were going to win. And now we see that that is exactly what happened with AWS, Google Cloud, and Microsoft Azure.
How to think about the application war?
When I think about how this application war is going to pan out I liken it to what we are seeing with startups. The vast majority of startups that are successful these days are fighting deep in the niches.
Why is this the case? Because the broad models have already been won by guys like Amazon and Microsoft and there are deep moats.
So if you wanna make a successful startup these days probably one of the best ways to do so is to find a niche with a big enough market that has not drawn a lot of attention from startups in the past.
And with the passing of years we see startups going deeper and deeper into lower layers of niches to find these opportunities.
AI applications will probably follow in a similar path
Right now for example startups are fighting on the first level of niches. For example maybe someone is fighting for the best AI application for lawyers or doctors with the goal that they become like a platform for those professions.
This is a lot like how startups thought 20 years ago. Who was going to be the de facto platform in the medical profession or for lawyers. And those battles were waged with some winners falling out.
In the coming couple of years the same thing will happen. The first layer of AI applications will shake out with the initial winners. And for example there will be the de facto lawyer AI co-pilot that thousands of lawyers will use throughout their day for all kinds of use cases.
But there are lots of use cases for a lawyer. And so some of those use cases will only be handled at a mediocre level.
So some other startup will come along and focus just on a few of those use cases where the defacto platforms are weak, but the market size is still interesting enough.
And a second layer of AI application startups will form.
This will continue in tons of industries until we see many thousands of successful AI applications.
Let’s take the example of DALL-E 3
DALL-E 3 is the AI image generator that was produced by OpenAI. I use it fairly frequently and it’s pretty good.
I can put in a prompt and it gives me a decent image.
And for my purposes it’s fine. If I want a slightly better image than I change my prompt.
But if i have a specific design use case that I want to create an image for.. than probably I need something a bit more specialized.
For example there are companies like Design.ai that are already attacking those use cases. And will create me images that are specifically tailored to things like logos or social media posts.
This is exactly what is happening right now in tons of niches. Just google search something like “AI for XYZ niche” and you will probably see some of these early players pop up.
For example this is what pops up when I searched for “Ai image generator for designers.”
How to leverage this?
So if you’ve understood what I wrote above you should be pretty excited in my view. Because there are going to be hundreds, if not thousands of AI application layer companies that are going to be successful in the coming years.
And that smells of opportunity!
Because while perhaps the SAAS models in your niche are already occupied and highly competitive… the AI copilot for your niche is probably not.
And guess what.. that those SAAS players will probably have to basically cannibalize their own revenues to become successful in AI. Which is something they will likely be hesitant to do.
How Incumbents will Approach AI: The Preply Example
An example I came across not long ago of this was the company, Preply. Preply is a marketplaces for language tutoring and it has achieved somewhere around a $500 million valuation, having raised around $175 million from VC’s.
Now they have a very difficult decision to make. Because there is a very high likelihood in my view that humans will be completely cut out of the language learning process due to AI in the future.
I mean… it’s just easy to envision a future where I turn on my AI language tutor and just talk to it anytime I want, day or night. And it tracks my exact progress, going over the exact things I need to get to a conversational level ASAP.
What value would a human add? If there is something, then I fail to see it.
But they are also naturally going to be concerned about this and so they are trying to incorporate AI into their business. But from my perspective they’re highly likely to do so with the wrong motivations.
Because the problem they are trying to solve is most likely not… “Give me the best AI-enabled solution for my language tutoring niche.”
Rather it is likely “I want to be the best AI-enabled solution for language tutoring that incorporates the many thousands of human language tutors I have.”
Because that is their moat and would be foolish of them to completely throw away the beautiful business they have built.
And that, in my view, is the achilles heal of all these incumbents jumping into AI.
Which is the weak point that you entrepreneurs out there need to jump all over.
