How to become a data business
The only way to become a data business; 8 lessons on building products with generative AI; The burden of knowledge is upon us, but not on AI
This is the Three Data Point Thursday making your business smarter with data & AI.
Want to share anything with me? Hit me up on Twitter @sbalnojan or Linkedin.
The only way to become a data business
8 lessons on building products with generative AI
The burden of knowledge is upon us, but not on AI
Let’s dive in!
8 lessons on building with gAI
A confession: I don’t do videos well. At least not the ones I have to record myself in.
That’s why I’m psyched to see Synthesia’s Series C. If you don’t know what they offer: AI avatars as substitute actors in your videos!
Check out their example section if you want to see them in action.
(Source: Product demo on YouTube)
So what? It’s not just Synthesia’s Series C that’s exciting but also what Matt Turck shared on building generative AI (gAI) data companies and Esteban Sosnik, who shared similar stuff a while back.
Here’s our synthesis of 8 great lessons:
Now is a great time to start a company for gAI infrastructure like models, data, or APIs.
If you’re a model user, you have to have a unique approach to model and or data. Otherwise, you’re a sitting duck for any incumbent with more money.
Go for solving bigger problems for customers instead of “efficiency gains.”
Leverage research to build a competitive moat. Like Seldon, they are in a good spot for that.
If you solve a bigger problem, provide a complete experience, not just the “fun AI parts.”
Example: Synthesia doesn’t just offer AI avatars; they give you the entire video editing studio + AI avatar.
Focus on solving business problems, not AI problems. Otherwise, you won’t last beyond the hype.
Example: Synthesia solves the enterprise problem of producing teaching videos for 1,000s employees—a series of enterprise-level problems.
If you can’t provide infrastructure or solve bigger problems, integrate deep into your existing product.
Don’t try to provide users with a new workflow; instead, supercharge the one you already have!
Example: pandasAi automatic generation of a heatmap. That was already possible before, but now it’s much easier!
For some great examples of generative AI integrated into products, check out this tweet:
AI & the burden of knowledge
TL;DR:
humans = the only species that pass knowledge on
passed on knowledge => knowledge accumulates
learning all knowledge => harder for each generation
=> tadaaa this is called "burden of knowledge"
So we are doomed or have to invent ways to learn knowledge faster and faster (hey computers! And brain-machine interfaces…).
…It’s like a game of Jenga, every turn moves the game further, but also makes it harder for the next player to move advance even further.
So what? AI…AI accumulates knowledge without the burden of knowledge (it cannot die, and can learn faster by adding more GPUs...).
The result: => We're in a weird, potentially screwed, but at the very least, unpredictable situation.
AI will leave us behind in knowledge, and we won't be able to comprehend how much.
Analogy: We will become monkeys looking at the night sky, wondering who turned on all those funky lights.
Why care? Because this will scare people! And it will turn into an unpredictable risk we all gotta account for. Whether you use it, or build it, this is an essential force in play.
How to become a data business
There’s almost a gazillion ways to use data.
But in our experience, only one that will turn you into a data business.
What’s data business? Data business = the Amazons, Netflixes, AirBnBs, and Spotifys of this world. The businesses that bring value to customers with the help of data.
Why care? We have a simple opinion: Every business has to become a data business or go under. There is no good reason to build any other startup or invest in new products that are not data-heavy.
Founding is easy, but where do you start as an established business?
The six typical use cases:
Improve business decisions
Understand customers
Create smart services
Create smart products
Improve processes
Monetize data itself
Here’s a secret: There is only one right choice of those six use cases, and almost every company chooses wrong.
What is the only one that should count for you? Smart products, smart products, and smart products.
So what? The thing is, most companies start with improving business decisions, and that’s almost always the wrong choice.
Why? Because all businesses already make decisions. The gain in using data there is small. But most products don’t use data or AI. The gain there is going to be big.
Think about Amazon. It’s first big data initiative? Introducing the “see what others also bought” drives 10-20% more revenue, thus scaling while the company grows.
Think LinkedIn and their People You May Know Feature (PYMK) (the first feature of such a kind at LKD) with a 30% higher CTR than any other prompt.
Spotify with their Discover Weekly and the viral Spotify Wrapped feature.
Netflix with their $1 million price (in 2009!) to improve their recommendation engine.
The evidence is clear; you should become a data business; And you should start with product, product and product, not with business decisions.