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
So AI is on the tip of everyone’s tongues these days…and it’s widely available to everyone.
Even more so now that GPT-4o is available completely free of charge by OpenAI. And so pretty much the entire world has access to it and many folks are using it a bit.
Plus there are a number of other options like Google’s Gemini and Claude AI by Anthropic that might be a bit better or a bit worse at certain things. But in my book you can almost say that the base availability of AI via a chatbot is commoditized.
It’s kinda like when I go to the shop and buy my eggs. I buy different brands all of the time without really paying attention. Sure some might taste a bit better and others a bit worse… but enough that I actually have a true brand preference.
Also the LLM space is tough to make money
A couple weeks back when ChatGPT decided to make its latest model, GPT-4o, completely free it quietly made commercial history in my view.
It was the first time a company took a product that it had millions of paying customers for and decided to make the latest version of its product absolutely free. Thus potentially wiping out many millions of revenue because all of the folks paying $20/month for premium didn’t really have any reason to pay anymore.
Why did OpenAI do this?
Simple. They don’t really need the money because they are so flush with investor cash that they are burning at a very high pace. And so they’d rather focus on user growth.
Couple this with the fact that compute costs are extremely high and it makes this space an extremely difficult one to make money.
Do you really wanna be the startup that creates its own LLM and competes against deep pocketed players like OpenAI who are willing to offer their best product completely free?
Probably not.
Ok so we’ve all got really good generative AI… so what happens now?
After all a lot of big names like Thomas Smale, CEO of FE International, have said things like “viewing AI not just as a tool for efficiency, but as a catalyst for strategic transformation. It opens doors to new markets and innovative business models.”
Well, the reality in my view is that the hype IS in fact being realized as we speak.
For example, startups are using AI as a core part of the product to build all kinds of great things. But the models they are using are the ones that everyone is using… that part truly is like a commodity.
Rather where they are differentiating are mainly in two areas:
- the proprietary data that they feed these models
- the user experience in how they solve the problem
Let’s now break down these two things.
Proprietary data that feed these models
Companies that have been in existence for awhile have collected tons of data. This data is not available to companies like OpenAI to train their models on this data.
And as a result things like ChatGPT are always going to remain a bit general. They just don’t have access to the data to be great at niches where there is a lot of non-public data.
For instance e-Commerce. Let’s say that Amazon decides it wants to create an AI that predicts exactly what you want based on all of the data that it knows about you.
Oh wait.. whoops! Amazon has already been doing that for many years and using it to decide what it shows you every time you visit their site…. lol
You see… ChatGPT is never going to have nearly as much data about your consumption patterns as Amazon is… and so they will never even play in that game.
Amazon is just one example… but you can think of equivalents in pretty much every industry and niche. The opportunities are just huge!
The user experience
Another place where startups are innovating is in the user experience. On certain types of problems a chat interface is just not the right way to solve the problem.
My Amazon example above is a perfect example of this. It is not that you need to enter a query on Amazon’s site and ask.. “Hey show me the 50 products that you think I want to purchase right now.”
That would be foolish because there is already an accepted way that you shop and purchase items on Amazon, and therefore they blend the AI into that.
The same will hold true for all kinds of problems.
A good example is this whole notion of ‘co-piloting’ that is already very popular among software developers.
You just do what you normally do when you are working and the whole time the AI is learning and trying to make you more efficient.
Eventually all of the tasks that you repeat frequently are just done for you.
The future for AI is exciting and the hype is real!
That is what I like to tell friends when we discuss AI. It’s not like other hype trends that haven’t panned out at scale yet (eg. Metaverse, augmented reality, etc.)
Rather AI is creating value in all kinds of industries already and growing at a lightning pace.
It truly is something you need to just continue to invest time in and stay on top of the latest developments. Because it will pave the way to massive efficiency improvements and new ways of working.
I’m excited for this future… and I hope you are too!