Low-code platforms promise to democratize software development: drag-and-drop interfaces, pre-built components, minimal programming. But the platforms' heavy reliance on graphical user interfaces introduces accessibility barriers for users with visual impairments, color blindness, and motor disabilities.
Analysis of user reviews for accessibility issues (arXiv:2504.19085) applies NLP to detect accessibility-related complaints in app store reviews, achieving 78% F1 score. The hybrid model combines keyword-based detection with contextual understanding to identify reviews mentioning screen reader incompatibility, color contrast failures, and touch target sizing problems.
The finding: accessibility issues in low-code applications are systematically underreported. Users with disabilities encounter barriers but often don't describe them in technical accessibility terms. A review saying “I can't read the buttons” might indicate a font size problem, a contrast problem, or a screen reader labeling problem. The NLP model must interpret user complaints through an accessibility lens to surface the underlying issues.
The deeper irony: platforms designed to make development accessible (to non-programmers) inadvertently make the resulting applications inaccessible (to users with disabilities). The democratization of one kind of access creates barriers to another kind. The drag-and-drop interface that empowers a non-coder produces output that excludes a blind user.
This suggests that accessibility should be a constraint at the platform level, not an afterthought at the application level. If the low-code platform enforces minimum contrast ratios, mandatory alt-text fields, and proper semantic structure in its components, every application built on it inherits accessibility. The platform is the leverage point — not the individual application developer who may not know accessibility guidelines exist.