
2026
New Mexico Department of Health: 600+ public health documents, from 2,000+ accessibility issues to zero
A parent filling in an asthma action plan for school, someone requesting a water analysis, a family reading about mpox in Diné: New Mexicans get much of what they need from their Department of Health as PDFs. We committed to 300 of the Department’s public documents and delivered more than 600 in one month (May 2026), taking them from more than 2,000 WCAG 2.1 AA issues to none.
- 2,000+ → 0
- WCAG 2.1 AA issues across the documents, before and after
- 600+
- public documents remediated
- 2×
- the 300 documents committed
- 1
- month, May 2026
Schematic — client UI under NDA
- Client
- New Mexico Department of Health
- Industry
- Public sector
- Scope
- Document inventory · Automated remediation · Human review · Verification · Audit trail
- Year
- 2026
Challenge
The New Mexico Department of Health publishes much of what people need on its website as PDFs. Forms to fill in, like the asthma action plan a parent gives a child’s school, a water analysis request, a report of audiological results or a clinic’s patient-navigation form. Signs and pamphlets, like the measles “Call before entering” poster and an mpox fact sheet in English, Spanish and Diné. And the decisions and protocols behind the Department’s programs.
For someone who uses a screen reader, a PDF works only when it carries structure: tags, a logical reading order, alternative text for images, headers for tables and, in a form, a name for every field it announces. Some of these forms have more than 150 fields. Across the documents, our checks found more than 2,000 WCAG 2.1 AA issues.
WCAG 2.1 AA is the standard the Department of Justice’s ADA Title II rule sets for state and local governments, and the rule covers electronic documents such as PDFs; for a state agency, the compliance date is April 26, 2027.
Approach
Every document went through the same steps, from inventory to delivery:
- Inventory: each PDF was checked against WCAG 2.1 AA with veraPDF and our own detectors before anything changed, so every issue started on the record.
- Automated fixes: mechanical issues, such as untagged content, missing structure and document metadata, were fixed automatically, and each fix was re-checked against the same rules.
- Human review: judgment calls went to people. Reading order, color contrast, alternative text and table summaries were decided by a remediator. AI running on our own servers drafted alternative text and table summaries; a person accepted, edited or rejected every draft.
- Verification and delivery: each document was checked by its remediator, then by a second person, and passed an independent check against WCAG 2.1 AA before delivery. A document was approved only when every check was resolved, with an accessibility report of what was tested and what changed.
- Audit trail: every fix, every review decision and every status change was recorded with who made it and when.
Ockem Proof, the platform we built for document remediation, was not ready yet. Some of its capabilities already were, and they carried the mechanical work, which is what let the team go past the commitment.
AI drafts. People decide. Everything leaves a trail.
Impact
The commitment was 300 documents. In one month, May 2026, we delivered more than 600, twice the volume, and the issues found against WCAG 2.1 AA went from more than 2,000 to none. A parent, a patient or a clinic that uses a screen reader can now read these documents in order and fill in every field, and each one comes with its own record of what was checked, what was fixed and who decided.
We have worked with the Department since 2025.


