Accessibility scanning, now enhanced with AI.
Every issue explained. Every fix suggested.
WCAGScan’s rule-based engine finds the accessibility barriers on your site and calculates your score. AI then helps you fix them, with a plain-English explanation and corrected code for each problem, ready to copy.
How it works
The scanner does the testing. AI does the explaining. You stay in control of every change.
-
01 · SCAN
Rules find the issues
Our Deep Scan engine renders each page in a real browser and checks it against WCAG 2.0, 2.1 and 2.2 success criteria: computed contrast, keyboard focus, reflow, structure and more. Your score comes from these checks alone.
-
02 · EXPLAIN
AI suggests the fix
Each distinct failing element is explained in plain English, with a corrected version of the code. Identical problems repeated across pages are grouped, so you fix the template once.
-
03 · FIX
You review and apply
Suggestions appear in an “AI Suggestions” tab in your interactive report and in an appendix to the PDF. Copy what you need, fill in any [bracketed] wording, and test.
Two things AI adds to your report
Suggested fixes
For the highest-priority issues (Level A first, then AA, most frequent first), you get:
- What is wrong with that element, and why it fails the criterion
- A corrected snippet of HTML or CSS, with a Copy button
- A note on how to verify the fix, or when another technique applies
- Colour fixes re-checked by WCAGScan’s own contrast calculation
Wording review
Some barriers are about meaning, which rule-based checks can’t judge. AI reviews the page titles, headings, link text and image alt text on your pages for:
- 2.4.2 Page Titled: vague or duplicated titles
- 2.4.6 Headings and Labels: headings that say nothing about their section
- 2.4.4 Link Purpose: ambiguous links like “Go” or a bare URL
- 1.1.1 Non-text Content: alt text that is a file name or just “image”
What it looks like
Two examples in the format your report uses. The code and text below are illustrative, not taken from a customer site.
SC 1.4.3 Contrast (Minimum) · Level AA · 14 instances
Scanner finding: Text contrast 2.54:1 is below the 4.5:1 minimum for normal text.
What to change: The light grey footer text is too faint against the white background for people with low vision to read. Darken the text colour while keeping the same grey tone.
.footer-note { color: #9ca3af; background-color: #ffffff; }
.footer-note { color: #6b7280; background-color: #ffffff; }
Checked by WCAGScan: suggested colours give 4.83:1 — meets 4.5:1
| Page | SC | Text | Problem | Suggestion |
|---|---|---|---|---|
| /products | 1.1.1 | IMG_2041.jpg | The alt text is a file name, so screen reader users hear nothing useful. | alt="[describe what the photo shows, e.g. the product in use]" |
| /services/hosting | 2.4.2 | Home | The title does not identify this page, and it is the same on 6 other pages. | Managed Hosting Plans | [Company name] |
| /blog | 2.4.4 | Go | The link text does not say where the link leads. | Read the full post: [post title] |
Built to be honest
AI can be wrong, so WCAGScan keeps it in an advisory role and checks its work where it can.
What AI does
- Explains each issue in plain English
- Suggests corrected code that keeps your original tags and attributes, changing only what the fix requires
- Marks wording only you can supply with [square brackets] instead of inventing it
- Prefers native HTML over ARIA, the approach WCAG guidance recommends
What AI never does
- Change your score, grade or issue list. Those come only from the rule-based scan
- Touch your website. It has no access to your site, browser or credentials
- Recommend overlays or scripts that alter your pages for users
- Claim compliance. Suggestions are not a certificate or legal advice
Checked, not just generated
Every suggested colour pair is re-measured with the same contrast maths the scanner uses, and the report says whether it passes. A wording finding is dropped unless the text it quotes really appears on that page.
Your data
AI sees only what the scan already collected from your public pages: the failing element snippets, the scanner’s findings, and each reviewed page’s title, headings, link text and alt text. Nothing else from your site is sent, and no login details are ever shared.
Questions
Does AI change my accessibility score?
No. Your score, grade and issue list come only from WCAGScan’s rule-based checks. AI suggestions sit in their own tab and PDF appendix and are never counted.
Will it fix my website automatically?
No, and that is deliberate. Automatic “fix” overlays often create new problems for assistive-technology users. WCAGScan gives your developer corrected code to review, apply in your own source and test.
Can the suggestions be wrong?
Yes, as with any AI. That is why every suggestion is labelled “review before use”, colour fixes are re-checked by the scanner, wording findings must quote real text from your page, and anything only you can know, such as what an image shows, is left as a [placeholder].
Which pages and issues get suggestions?
Fix suggestions cover the highest-priority distinct issues first (Level A, then AA, most frequent first). The wording review covers the first pages of each scan. Your report says how many were included.
Which AI model do you use?
WCAGScan uses Claude, a model from Anthropic, through Anthropic’s commercial API. The model name is printed at the bottom of the AI Suggestions section of each report.
Is it included in my plan?
Yes. AI-assisted suggestions are currently included with every scan, including the free scan, at no extra cost.