Schmidt argues the vertical app’s data moat comes from two stacked flywheels: an across-customer one (patterns that compound as you see more variants of the same problem) and a within-customer one (the why behind specific decisions, the unsaid exceptions, a firm’s own rules of thumb that surface only through real interaction). None of this is on the public web, so “no amount of training compute substitutes for being inside the workflows where this knowledge actually lives.”
The non-obvious point is the mechanism: a horizontal agent could in principle build the same learning infrastructure — the reason it doesn’t, beyond focus, is UX. “Capturing this kind of knowledge depends entirely on the workflow surfaces you give the user, and vertical players can shape those surfaces around exactly what their workflow needs to surface. Horizontal tools can’t.” This is the engine behind Proprietary feedback loops create moats that widen with every interaction, it explains why Tribal knowledge is the irreducible human input that enables agent automation is capturable only by focused players, and it is the source of the compounding in The context flywheel is a Day 90 moat — Day 0 comparisons are misleading and the broader Context is the product, not the model thesis.