UXDX Community:The AI Edge: Prototype Real, Ship Smarter

In B2B logistics software, workflows are data-heavy and mission-critical. Each customer has their own specific workflows and data variations. Traditional prototypes with placeholder data can't capture this reality—and that creates a blind spot. Users can't give meaningful feedback until they see how a solution handles their actual data, edge cases, and terminology. This delays the conversations that matter most.
We started experimenting with AI-assisted prototyping to change this. Using tools like Cursor, we built prototypes with much higher data and behavior fidelity during discovery—something that would typically require developer time we didn't have. The AI handled the complexity, letting us work with real customer data before any engineering resources were committed.
I'll share how we integrated AI into our discovery process and what we're learning—how increased fidelity shifted our focus to the problems that actually matter earlier in the process, how it changed the types of conversations we had with users and stakeholders, and how it enabled earlier collaboration between UX, product, and engineering. I'll also cover where this approach helped (and where it complicated things).

AI is changing front-end work, but what matters most to product teams is where it actually improves outcomes for users and teams. In this talk, I share personal case studies from building on large web platforms, focusing on real decisions, tradeoffs, and results rather than AI hype. Through practical examples, I show where AI helped with interface design, accessibility reviews, component scaffolding, and understanding UI telemetry, and where it introduced friction or false confidence. The goal is to give a UXDX audience concrete lessons they can apply across design, engineering, and product collaboration to ship better experiences faster without lowering standards for usability, performance, or accessibility.