PDF-to-HTML AI Conversion Service
Convert PDFs to Accessible HTML
NCDIT, in collaboration with Google, has developed an AI-powered PDF-to-HTML conversion service using Gemini AI. The service helps state agencies transform PDF documents into more accessible, editable web content while reducing the time and effort required for manual conversion.
Converting PDFs to structured HTML supports digital accessibility, improves the user experience, and makes government information easier to find, maintain, and use across devices and emerging technologies.
FAQs
HTML should be the preferred format for most website content. Unlike static PDFs, HTML pages are easier to update, adapt to different screen sizes, and work with assistive technologies.
- Responsive across devices
HTML content automatically adapts to desktop, tablet, and mobile screens, making information easier to read and navigate. - Improved accessibility
Properly structured HTML supports screen readers, keyboard navigation, and other assistive technologies, helping agencies meet WCAG accessibility requirements. - Easier to find and search
HTML content can be indexed by search engines and searched directly on websites, improving discoverability and supporting more useful analytics. - Supports translation
HTML content works more effectively with language translation tools, helping agencies reach a broader audience. - Easier to maintain and update
Content can be edited directly within a website without requiring changes to a separate PDF document. - Ready for emerging technologies
Structured HTML makes information easier for AI-powered search, language tools, and other technologies to interpret and use.
The service uses Gemini AI to analyze PDF documents, identify their content and structure, and generate accessible HTML that can be published as a webpage.
- Submit PDF URLs
Agencies provide a CSV file containing the URLs of the PDF documents they wish to convert. - AI analyzes and converts the content
The service extracts text, images, tables, and document structure. It also attempts to:- Generate alternative text for meaningful images.
- Identify decorative images.
- Remove unnecessary page breaks, page numbers, and repeated headers and footers.
- Convert document headings, lists, and tables into structured HTML.
- Receive converted HTML and review reports
Each converted document includes three deliverables:- WCAG 2.1 Level AA-targeted HTML version: A structured HTML version designed to meet accessibility requirements.
- AI accessibility audit: A report identifying potential accessibility issues that may require human review.
- PDF-to-HTML fidelity report: A comparison of the original PDF and converted HTML to help verify content accuracy and completeness.
- Review, edit, and publish
The web manager reviews the generated HTML and accompanying reports, makes any necessary corrections, and chooses whether to finalize and publish the HTML page.
Important: Although the service generates HTML designed to meet WCAG 2.1 Level AA, automated conversion does not guarantee full accessibility compliance. Agencies should review the generated content and reports before publishing.
The service prioritizes content accuracy and accessibility rather than reproducing the original PDF's visual appearance.
- Formatting will change. Fonts, colors, spacing, and layout may differ from the original document. Published HTML will use the styles of the destination website.
- Scanned documents may require corrections. The service uses Optical Character Recognition (OCR) to extract text from scanned PDFs. Results depend on scan quality, and unclear or low-resolution documents may require manual corrections.
- An auditor report is provided. The report helps identify areas for review but does not replace human accessibility testing.
Only PDF files are supported. Other document formats cannot be uploaded directly to this service.
The service does not currently support the following types of PDFs:
- PDF forms, including fillable forms.
- Password-protected PDFs.
- Presentation slides saved as PDFs.
- PDFs consisting of a single image.
These limitations apply to this conversion service and do not necessarily mean the content cannot be made accessible through another method.
Although HTML is recommended for most website content, some documents may need to remain in PDF format.
| Situation | Recommended approach |
|---|---|
| A document must be available for offline use | Retain a downloadable PDF when necessary |
| A form must be printed, completed by hand, and mailed | Retain the PDF form and ensure it is accessible |
| A publication requires an exact visual layout, such as a magazine or formal report | Retain the PDF when the fixed layout is essential |
| A document is password-protected | Handle separately; the conversion service cannot process it |
When a form does not require printing, handwritten completion, or mailing, agencies should consider replacing the PDF with an accessible online form.
Important: Retaining a PDF does not automatically exempt it from accessibility requirements. Agencies should evaluate applicable Title II requirements and exceptions and ensure documents that remain in PDF format are accessible when required.
Pre-publication review checklist
AI-assisted conversion can significantly reduce manual work, but it cannot guarantee full WCAG conformance. Every converted page should be reviewed for content accuracy, formatting, and accessibility before publication.
Use the following checklist to review the generated HTML.
- Headings follow a logical hierarchy and accurately describe each section.
- Bulleted and numbered lists are correctly structured.
- Images have accurate, meaningful alternative text, and decorative images are appropriately marked.
- Tables use proper HTML structure, including identified header rows or columns.
- If a table appears as both HTML and an image, remove the image copy and retain the accessible HTML table.
- Hyperlinks use descriptive text that makes sense outside the surrounding content.
- Text extracted through OCR is accurate, complete, and free of recognition errors.
- Unnecessary page numbers, repeated headers, and footers have been removed.
- Content is complete, correctly ordered, and readable on desktop and mobile devices.
- The auditor report has been reviewed and identified issues have been addressed.
The PDF-to-HTML conversion service is designed to help agencies reduce document remediation efforts and support compliance with ADA Title II and WCAG accessibility standards. However, AI-generated content still requires human oversight.
Agencies remain responsible for reviewing converted content, correcting identified issues, and ensuring published information meets applicable accessibility requirements.
There is currently no cost to state agencies for using the PDF-to-HTML conversion service.
How to Access the Conversion Service
Agencies interested in using the PDF-to-HTML conversion service can submit a request through the NCDIT Support Center.
As part of the request, agencies should prepare a CSV file containing the URLs of the PDF documents they wish to convert.