Designing with Intent: A New Layer for AI-Generated UI

AI design agents can now turn plain-language prompts into fully functioning UI flows in seconds. This is a genuine leap forward—and a genuine risk. When anyone can generate a cancellation flow, checkout, or onboarding sequence from a simple description, a critical question often gets skipped: what is this flow actually trying to do to the user? This talk explores how to close that gap. What if we capture a flow’s intent beforeanything is generated—articulating the business goal, the user’s goal, and what success truly looks like from the user’s perspective? And once that intent is defined,what becomes possible? We could flag conflicts between business and user interests, surface likely dark patterns, and generate constraints that persist throughout the AI design pipeline. At the center of this approach is the intent interface—a structured layer that makes implicit design decisions explicit, and therefore auditable. The core insight is simple: dark patterns aren’t a code problem; they’re a design problem. And as AI agents increasingly generate UI at the design layer, the point of intervention must shift upstream—to the moment intent is defined, before a single pixel exists. You’ll leave with a new mental model for where ethical “linting” belongs in an AI-assisted design workflow—and the architectural questions your team should be asking before shipping another AI-generated experience.