LLMs and foundation models are all the rage right now, and understandably so. They are powerful tools that can be used to build a wide variety of applications, and they have the potential to revolutionize many industries.
As is the case with most frontier technologies, the ones building the strongest products on top of it are those who are building the entire vertical. With the AI boom, this vertical requires an insane amount of resources to build and maintain, which is why we see so many startups trying to build quick wrappers around these models to capture some of the market share.
They are easy to dismiss as “nonsense products”, but I think they serve a purpose in the grand scheme of things - for the innovative bunch. They are built to grab attention and capital, which can then be funneled into more meaningful and innovative projects.
I’d expect, past this initial stretch, foundation models to be a not-so-innovative commodity that is taken as a given the same way we treat hosting, or cloud compute right now.
Sure, current major AI companies provide both the platform and the product, but when (and if) we reach the development plateau for the underlying platform, then it will be those providing the correct tweaks/incremental improvements who will prevail as customer favorites.
I agree with those online that most of the wrappers striving for revenue-optimization over sustained PMF will be gone one way or the other as the base technology settles.
In my humble opinion the smart founder of these companies will do one of the two things till then:
- Option 1: Use the initial penetration of the easy product and capital acquired during this boom to quickly pivot and generate a resource-heavy USP in a vertical that creates separation (i.e. build a company, not a product)
- Option 2: Take the earliest check that makes them happy, and move on
The odd founder (exceptional or the exact opposite) will hold onto the product and potentially make something cool out of it.
Ever the example: Instagram and TikTok didn’t necessarily kill Snapchat, but they sure did not help its case either. Regardless, it managed to survive and thrive in its own niche.
Personally, I’m a big fan of building companies that matter and not clinging onto most products too tightly.
If the means to starting that company off is to build a few not-so-groundbreaking products that payroll actual innovation later on, I’m all for it.
They are the companies I believe can create the right conditions for good progress. They’re the ones I’d bet on.
For all my rationale, there will always be exceptional cases, fueled by strong convictions about what a product can be, and driven by the founders’ desire to prove those calling it “nonsense” wrong.
Those are the ones that create great progress. Those are the ones I root for.