As AI-powered analysis, accessibility tooling, and machine-readable disclosures become increasingly important, it may be worth considering enhancements to the semantic HTML generated in EDGAR outputs.

Currently, document headings can be visually formatted but may not be represented as semantic HTML heading tags (<h1> through <h6>) in the final output. Similarly, table header rows may not be represented as HTML header cells (<th>), even when they are clearly intended to function as headers.

While these approaches have no impact on how filings render visually, semantic markup provides additional structural information that can help:

• AI tools better understand document hierarchy and section boundaries
• Improve context retention when content is extracted or segmented
• Strengthen relationships between table data and their associated labels
• Enhance compatibility with accessibility technologies
• Improve machine readability without affecting filing appearance

A potential enhancement could be the ability for EDGAR exports to:

• Map Heading 1, Heading 2, Heading 3, etc., to corresponding HTML heading tags where appropriate.
• Export designated table header rows and columns using semantic header cell markup.
• Preserve semantic structure while maintaining the same visual presentation in the rendered filing.
• Provide workspace or document-level settings to control semantic markup behavior.

As AI consumption of public filings continues to grow, support for additional semantic markup could provide meaningful value without requiring changes to document appearance or user authoring practices.

Interested to hear whether others see value in enhanced semantic HTML support as part of future EDGAR output capabilities.