Articles in
components
We audited seven React component libraries shipping in 2026 — Radix UI, Headless UI, shadcn/ui, Mantine, Chakra UI v3, Ark UI, and React Aria Components — scoring each on axe-pass rate, ARIA pattern coverage, keyboard contract, and bundle-size cost.
data-viz
An engineering primer scoring Vega-Lite, Plotly, Observable Plot, Apache ECharts, and D3 against accessibility defaults — SVG/ARIA, colour-blind palettes, keyboard-navigable data points, screen-reader hierarchy, and alternative table view — with concrete picks by use case.
ai
An engineering primer on the state of AI-generated alt text in 2026. We tested GPT-4o, Claude 3.7 Sonnet, Gemini 2.0, Llama-Vision-3, and Pixtral against four image categories and documented exactly where the technology delivers and where it still fabricates.
aria
We tested aria-live regions across React 19, Vue 3.5, Svelte 5, and SolidJS 2.0 — four canonical patterns, three screen readers, every framework-specific gotcha. Here is the behavior matrix, the good-vs-bad code, and the playbook.
figma
We audited 50 production Figma files — anonymised, with permission — for the accessibility specs that did and did not make it into the handoff.
eye-tracking
How modern web apps work — and break — for users navigating via eye-tracker, head-pointer, or switch input. A concept primer on the hardware, the relevant WCAG criteria, and the design patterns that survive single-axis input.
generative-ai
Writing system prompts that make ChatGPT, Claude, Gemini and Be My AI useful to screen-reader users is becoming its own design craft — with rules about structure, em-dashes, AT handoff, and the open UX problems nobody has solved yet.
ios
A head-to-head primer on iOS UIAccessibility, Android AccessibilityNodeInfo, and the cross-platform bridges that try to reconcile them — what maps cleanly, what doesn't, and where the mobile web fits in.
pwa
Where progressive web apps stand on accessibility in 2026 — install-prompt UX, adaptive icons, screen-reader handoff between web and native, the file_handlers / share_target / window_controls_overlay manifest surface, offline assistive-tech behavior, and the iOS Safari install path post-iOS 16.4.
screen-readers
Four screen readers shape almost the entire assistive-tech web.
voice-ui
We benchmarked the four major voice assistants on Apple's Speech Accessibility Project and Google's Project Euphonia datasets — word error rate and intent-recognition rate by speech condition. Here is the matrix, the personalisation features that move the numbers, and what designers should ship.