Intelligence itself will be commoditized. Any generation of large models will have multiple competing versions from different labs. What’s going to matter is vertically tailored applications of artificial intelligence.
When founders approach me to discuss their ideas, I urge them to determine what matters for their particular vertical: data or process. Will the dominant competitive advantage be what you know or what you can do?
Incumbents will win most vertical plays that depend on data. They know more than any upstart. Even if they are slow to adopt AI, they’ll eventually overwhelm the competition with their ability to train or fine-tune models on their proprietary corpus.
Startups can win where process matters. Incumbents are at a disadvantage there. Their bigness—bureaucracy, complexity, policies, governance, risk mitigation—cannot easily be changed or reinvented. Startups can do things much better from scratch. It just has to be a space where the doing is what matters.
Understanding a process is a type of knowledge, but it’s the exact type of tacit intelligence that AIs are bad at. The best places to find and create durable value using AI will be where the secret sauce is in the little details, the workflows, the division of labor between agents, and the human taste that improves the experience.