AI Did Not Begin in 2022
The breakthroughs we celebrate today sit on top of decades of applied work that most people never noticed.
The Hype Cycle vs. The Scaffolding
Generative AI has reached the point where the spectacle is giving way to something more useful. That is good. What is not so good is the way this shift has flattened the entire history of AI into one story about large language models. It is an understandable story, but it is also wrong. It erases decades of applied AI work that has been solving hard, vertical problems in medical imaging, legal expert systems, strategy tools, and more (hello? Quant Trading?). The breakthroughs we see now did not appear out of thin air. They sit on top of generations of builders who kept working long before anyone cared.
The Illusion of Expertise
The current wave of generative AI feels magical, and the sense of wonder is real. But magic is also an illusion. What we are seeing is advanced automation that produces the appearance of expertise for people who have only just encountered it.
It lowers barriers, which is great. It also hides the depth of work required to build something durable. This is why the market often confuses two completely different things.
Novelty: A model that can hold a convincing conversation or generate a draft.
Value: A system that creates a structured workflow and produces a reliable and valuable outcome.
One is amusement. The other is a product.
VSTRAT’s Moat: Architecture Is the New Algorithm
VSTRAT fits into the long lineage of applied AI. The work that led to it started years before we ever set foot at INSEAD and is the reason we were invited. We arrived in 2014 as an obscure research project and learned quickly that the real value in strategy systems comes from architecture, workflow, and data design, not from raw AI power.
This has been true for decades. It is still true now.
Our moat is simple. The magic is not the chattiness of the algorithm. The magic is the application, the value it creates, and the long history of exploring and learning what works and what doesn’t.
We focus on structure rather than spectacle. Instead of letting users bounce around in an unstructured chatbot, we enforce the discipline needed for real strategic work.
We focus on scalability rather than service. A YAML-defined system can scale instantly while staying consistent. That is what makes the product reliable, and reliability is what schools and companies are actually buying.
The Practitioner’s Truth
Applied AI practitioners have known for a long time that sustainable value has nothing to do with headlines or viral demos. It has everything to do with solving actual problems.
Vertical applications often deliver more measurable value than broad general models because they understand the domain, the workflow, the data, and the user. That knowledge is not a relic. It is what keeps the field grounded.
As AI becomes more common, these builders are not an earlier generation the industry has outgrown. They are the ones who already lived the headaches newcomers are about to discover for themselves. They know that technology is judged by the value it delivers, not by what it promises.
The applied AI community has been proving that for years. The hype will eventually catch up to the substance. It always does.