Bridging the Screen Reader UX Gap in Web Forms Using AI-Assisted Step-by-Step Interaction

Many web forms can be technically accessible yet still frustrating to complete with a screen reader. The UX gap shows up in everyday moments: long sequential navigation, unclear structure, losing context across sections, and error messages that are hard to find and fix. This talk presents an AI-assisted, step-by-step interaction model that guides users through complex forms in smaller, manageable steps while keeping the experience predictable and user-controlled. I’ll demonstrate the approach with NVDA on a real form workflow, comparing the traditional experience against the guided flow and highlighting what improves; task completion time, error rate, and correction effort. You’ll also get practical engineering guidance on how to build it: progressive disclosure, correct focus management, accessible announcements, and safe boundaries for AI assistance so it supports clarity without changing the meaning of the form or taking decisions away from the user.