This week my AI Agent stopped working. Not because the model was broken. Not because the website project failed. Not because the dashboards disappeared. Not because the underlying research was lost. A simple billing issue somewhere in the AI infrastructure brought everything to a halt.
It was a useful reminder.
Many people think the AI model is the product.
I increasingly think they're wrong.
The model is becoming a commodity. Today it might be Claude. Tomorrow GPT. Next year it could be a model that doesn't even exist yet.
The real value sits elsewhere. The value is in:
- The data you have collected.
- The workflows you have built.
- The questions you ask.
- The judgement you apply.
- The intellectual framework you develop over years.
In my own case, I have spent decades studying macroeconomics, markets, demographics, energy, debt cycles, technology and digital assets.
The AI helps me organise, challenge and communicate those ideas. But the ideas themselves do not belong to the model.
This is an important lesson for investors as well.
Many companies are currently being valued as though access to a particular AI model creates a durable competitive advantage. I am sceptical.
Models will improve. Costs will fall. Competition will increase. The gap between today's frontier model and tomorrow's frontier model will narrow faster than many expect.
What will remain valuable is proprietary data, distribution, customer relationships, brand, trust and the ability to integrate AI into real-world workflows.
The winners may not be those with the best model. They may be those with the best systems built around the models.
In investing we often confuse the tool with the business. AI is no different.
The model is the engine. The business is everything built around it.
As investors, entrepreneurs and business leaders, we should spend less time obsessing over which model is currently winning the benchmark wars and more time thinking about how AI changes the economics of our industries.
The companies that prosper over the next decade may not be those building the most powerful models. They may be those creating the most valuable ecosystems around them.
The model is important. The system is what matters.