Why AI tools alone won't make companies identical

The piece challenges a common assumption about AI commoditization: that universal access to the same foundation models and tools should produce homogeneous products and strategies. The author's counterargument rests on three interconnected observations. First, differentiation occurs upstream of AI — in problem selection and framing. Most companies using Claude or ChatGPT are answering questions posed to them, but the competitive advantage belongs to those who notice which problems matter and frame them correctly. Second, AI has inherent limitations in complex, real-world domains. Training on the internet cannot capture the specific hardware in a hospital, the political constraints of a particular IT department, or the unstated preferences of a stakeholder. Companies that accumulate this tacit, local knowledge retain an edge no model can replicate. Third, AI can paradoxically weaken organizational decision-making by automating away the friction and disagreement that improve strategy. If every prompt builds on the last one's assumptions, users risk confirmation spirals that push bad ideas deeper — a risk traditional processes mitigate through human challenge and debate. The author implies that as AI becomes ubiquitous, competitive advantage will shift from "better AI access" to "better problem definition," "better local knowledge integration," and "better organizational friction management."

The author signals a move toward technical depth in future posts, indicating that while AI models converge, the architectural, operational, and strategic choices around how to deploy and integrate them remain the true differentiators.