Automated Accessibility Tester

Status: Candidate Project
Market Size: $848M (2026) → ~$1B+ (2034) at 12% CAGR
Last Updated: 2026-05-02

Overview

An AI-native web accessibility testing and remediation platform that moves beyond detection to automatic code-level fixes. Today's tools (axe-core, Pa11y, WAVE) excel at flagging violations but leave remediation to developers. This project generates actual, deployable code patches:

Automated fix generation: Generate HTML/ARIA/CSS patches for violations, not just detection
Crossing the 57% ceiling: LLMs can evaluate cognitive accessibility, reading clarity, and ARIA semantics in context—areas where rule engines fail
Intelligent prioritization: Rank violations by user-impact severity; group related issues; suggest remediation roadmap by business priority
Dynamic SPA-aware scanning: Test React, Vue, Angular single-page apps with runtime-aware component analysis

The Market Gap

The accessibility testing market is growing at 12% CAGR. Market drivers include:

ADA Title II ruling (April 2024): Public entities must comply with WCAG 2.1 AA by April 2026 (large) or April 2027 (small)
EU Accessibility Act (June 2025): Mandates accessibility for private-sector digital products across EU
Section 508 / WCAG updates: Compliance requirements increasingly codified as law

But all existing tools share a critical limitation:

Automated detection ceiling: Rule-based engines (axe-core, Pa11y) detect only 25–57% of WCAG violations
No fix generation: Tools flag issues but leave remediation to developers (manual, slow, error-prone)
Overlay stigma: JavaScript overlays (accessiBe, UserWay) are actively opposed by disability advocates and facing FTC enforcement
Unserved segment: No open-source tool offers AI-driven automatic remediation

Core Features

MVP (Must-Have)

Automated scanning engine built on or compatible with axe-core (MPL-2.0), covering WCAG 2.1 AA and 2.2 A/AA rules
Playwright and Cypress integration for CI/CD pipeline use; JSON output
Severity-ranked violation report with human-readable descriptions and WCAG success criterion references
SPA/dynamic-content support via controlled browser automation (post-hydration DOM testing)
AI-generated, context-aware code fix suggestions for at least the top 10 most common violation types (missing alt text, form labels, heading hierarchy, language attribute, button names, link text, colour contrast flags)

Should-Have (v1.1)

Natural-language explanation layer: For each violation, generate plain-English description of the problem and its user impact
Violation prioritisation and grouping: Cluster related issues; rank by impact across user disability personas; generate suggested remediation sequence
Monitoring mode: scheduled scans with trend reporting and regression alerting
WCAG criteria coverage reporting: tell users what percentage of WCAG 2.2 the tool covers vs. what requires human review

Nice-to-Have (Backlog)

LLM-powered assessment of previously manual criteria: reading level (3.1.5), focus-order logic (2.4.3), ARIA semantics in context
IDE extension (VS Code) for shift-left detection during authoring
Remediation dashboard with issue lifecycle tracking (open, in-progress, resolved, verified)
Export formats for legal/audit use: VPAT-compatible output, accessibility conformance report generation

AI-Native Opportunities

Fix generation, not just detection
- Current tools flag issues but leave remediation to developers
- An LLM could generate specific, context-aware code patches: corrected HTML/ARIA attributes, alt text from image context, focus management scripts
- 2026 arXiv preprint (López et al.) demonstrates this is feasible with LLMs on real Angular SPAs
Crossing the 57% automated detection ceiling
- Rule-based engines cannot assess cognitive load, reading-level clarity, or screen-reader UX flow
- LLMs that understand natural language and page semantics can evaluate criteria previously requiring human testers
- Examples: WCAG 3.1 (language/labels/instructions), 2.4.3 (focus order logic), meaningful ARIA usage in context
Intelligent prioritization and triage
- Existing scanners produce large, undifferentiated lists of violations
- An AI layer could rank issues by user-impact severity, group related violations, map them to personas (low-vision, motor-impaired, cognitive), and generate remediation roadmap ordered by business priority
Dynamic / SPA-aware scanning
- Most scanners still struggle with React, Vue, Angular where content is rendered dynamically post-hydration
- An AI agent understanding component trees and interaction patterns could test accessibility in a runtime-aware manner
Open-source differentiation vs. overlays
- Existing open-source tools (axe-core, Pa11y) provide detection infrastructure but no AI layer
- A free, open-source AI-native tester that generates real remediation code aligns with how developers actually want to work—unlike the opaque JavaScript overlays sold by commercial vendors (which disability advocates actively oppose)

Competitive Landscape

Tool	Type	Detects % WCAG	Fix Generation	Cost	Approach
axe-core	OSS engine	57%	❌	Free	Rule-based
Pa11y	OSS CLI	30–40%	❌	Free	Rule-based
WAVE	Free extension	~40%	❌	Free	Rule-based
axe DevTools	Commercial	57%	❌	$10K–$50K/yr	Rule-based
Siteimprove	Commercial SaaS	~60%	✓ (AI Q&A)	$10K–$30K/yr	AI-assisted
accessiBe	Overlay (FTC banned)	~50%	✗ (overlay, not fixes)	$49–$199/mo	JavaScript overlay
This Project	OSS + SaaS	57%+ → 80%+	✓ (AI-generated)	Free self-hosted	AI-native fix generation

Technical Design Considerations

Detection: Build on axe-core (MPL-2.0); add LLM-based assessment for manual criteria
Fix generation: Claude API for generating context-aware HTML/ARIA/CSS patches; integrate into code review workflows
SPA testing: Playwright or Cypress for post-hydration DOM analysis; wait-for-stable-DOM heuristics
Prioritization: Severity scoring based on WCAG success criterion weight + estimated user impact
CI/CD: GitHub Actions / GitLab CI native integration; PR comments with fix suggestions
Explanation: Plain-English descriptions of each violation; reference to WCAG success criteria; suggested user impact

Market Validation

Market drivers:
- ADA compliance deadline (April 2026–2027) creates immediate demand
- EU Accessibility Act enforcement (June 2025+) expanding compliance scope globally
- 4,000+/year US web accessibility lawsuits (2024 data) — companies actively seeking proactive solutions
- Disability advocates' FTC victory against accessiBe (April 2025) discrediting overlay approach
Customer personas:
- Enterprise development teams at regulated orgs (finance, government, healthcare)
- Digital accessibility officers / compliance counsel facing April 2026/2027 deadlines
- Web agencies and freelancers building accessible sites for clients
- E-commerce and SaaS companies facing ADA litigation risk

Why Build This

Market timing: ADA Title II deadline (April 2026–2027) creates immediate compliance pressure; EU EAA enforcement starting 2025
Technology maturity: LLM-based accessibility remediation papers (2025–2026) demonstrate feasibility; not speculative
Open-source gap: axe-core provides detection; no open-source AI layer for fix generation or prioritization
Regulatory advantage: FTC's accessiBe enforcement (April 2025) discredits overlay approach, creating demand for legitimate remediation tools
Accessibility ethics: Aligns with W3C WAI advocacy for source-code fixes over JavaScript overlays

Success Metrics

Adoption: 500+ GitHub stars within 12 months; featured as alternative to Siteimprove/axe DevTools
Compliance: Help 50+ organisations meet April 2026/2027 ADA/EAA deadlines
Fix quality: Generate deployable patches for 80%+ of top 10 violation types; user-tested patch accuracy >90%
Community: Active issue triage; 2+ contributors beyond core team

Resources: