Campus AI workflow
PilotFinds the document-accessibility problems software can catch reliably, and hands a qualified remediator the evidence for the rest.
- 17+Automated checks9 categories
- HumanFinal reviewRequired for every document
- ReviewableEvidence packageFindings and remaining work
Accountability
How this use case is governed
- Service owner
- TritonAI solutions team and campus accessibility partners
- Human oversight
- A qualified human validates reading order, semantics, alternatives, and the final accessible document.
- Measurement plan
- Issues detected, remediation time, residual accessibility findings, and reviewer agreement.
- Data boundary
- Public description; document handling depends on source data
Product media
See the workflow in action
Silent screen recording showing a PDF upload, automated accessibility findings, remediation progress, and the reviewable results.
- veraPDF Validation
- PDF/UA Standard
- Evidence Packs
How it works
How the workflow fits
The problem it addresses, what it actually does, and how far along it is.
Problem
Some of PDF accessibility work is mechanical and checkable. The rest depends on visual and semantic judgments software cannot safely make on its own.
Solution
The workflow looks for common structural and metadata problems, proposes fixes for some of them, and assembles an evidence package a human remediator can work from. It runs 17+ automated checks across 9 categories, covering reading order, tagged content, alternative text, and document structure, so a reviewer can see what was checked and what is still open. It assists expert work. It does not certify conformance.
Current status
Pilot. A finished document still needs a manual accessibility review against the appropriate standards. A staff member with no engineering background built this through the Citizen Developer Program. It runs in the browser, uses no database, and is ready for campus SSO.