While writing The Water Carriers, I constantly tried to imagine how AI would work in the future. I decided this experience would be highly personal and everpresent. It wouldn’t be a tool you ask questions, like today. It would be more like a companion always there to help you get the most out of life. When I saw what ChatGPT could do in 2023, I realized I had completely underestimated the future of AI.
I was already 70% through my book, but it was clear I needed to go back and rethink all references to AI. A key scene was in the third chapter. I wrote about the life of the uber-wealthy protagonist, Kasemchai, who invested in new apps. I originally speculated that apps would be so simple to build that teams wouldn’t pitch ideas with rough concepts. Instead, they would present fully functioning apps designed by behavioral scientists and emotion design specialists.
After seeing how AI was writing code, I realized that no one would code in the future. It would not only be cost-prohibitive and slow, but it would also be impossible even to understand what AI was doing with code. Kasemchai was investing in a dating app, but in a world where anyone could create any app they wanted, why would anyone invest in a company like this? If every kid could design a dating app and share it with friends, then App stores would be filled with tens of thousands of dating apps.
I had to rethink the value an app could bring, which made me consider how we would work with AI in the future. While we think today’s large language models (LLMs) are revolutionary, they will eventually become indistinguishable and a commodity. So, how will we build value in this world?
I started with brands like Amazon.com or Alaskaair.com to imagine what it would be like if anyone could create their own unique experience to interact with these brands. If you find a fun way to search for your next trip or buy shoes or appliances for your home, you would want to share that experience with your friends. The data for shopping would be valuable but available to anyone. And how would customers be protected in a world where anyone could create a clever or horrible shopping experience?
That’s when I realized the true value was not in the app or the UI but in the data—more precisely, the control of data that would allow privacy and security. This would allow you to create your own bank app that could securely access your bank account and transfer funds.
Therein lies the challenge. Anyone creating their own bank app could easily make mistakes tracking payments or fees and easily put their account in the red. And in a disseminated dating app experience, who owns your dating profile? How would people agree to make their profile visible in someone else’s AI-made dating app?
In time, I came up with this model:

Like your car or any technology, few people really know what’s going on under the hood. In this complex world, people are only aware of the AI Service layer (AISL). People choose these brands because they align with their identity. The AISL could be a large company, a religion, or a community.
While a religious AISL has access to all the LLMs available in the AI Personalization layer (AIPL), they would only present information aligned with the values or expectations of the AISL community. A less restrictive AISL could show all information to its users, even information that is dangerous or violates country laws. This is why a Compliance and Regulation layer (CRL) needs to be present.
The Data Integration layer would allow any country’s SPL to securely integrate with data anywhere in the world. Anyone could create their own bank or shopping experience, and banks and stores would know they are protected working with anyone, whether they design a unique visual experience or ask their AI companion for their bank balance. A bank would never send data to an AISL that didn’t comply with the CRL.
If you were to find a neat way to search for trips or track your expenses, how could you share that experience with other people in your AISL? There wouldn’t be a concept of apps. Instead, people could share a combination of preferences with experience-centric design or a PrefX. People could share their PrefXs on an AISL, and PrefXs could be monetized to allow anyone to make money from a clever AI-designed interface and experience.
How would a dating app work if there’s no vendor with your dating profile? That’s the beauty of this model. Your personal data would still flow up and down the layers. This would prevent underage children from participating in a dating app. It would protect people who want to overshare by following guidelines set by either a country or an AISL.
In this world, people interact with their companions so much that they lose sight of the AISL. Only when they meet a spouse or make a life change, do they change their AISL. That is where Kasemchai’s opportunity for enormous wealth opens up. Find out what he does in chapter three of my book when it comes out.