PDF accessibility remediation for institutions that have to prove it. Foliowright re-tags the PDF you already have — it never rebuilds the page, so the document still looks like the document — and writes a conforming PDF/UA structure into the same file. What it cannot determine, it refuses to guess and hands to a person instead, with the page and the image attached.
Six of the hardest real PDFs we could find, published by six US universities — 419 pages with 46,699 machine-detected failures as published. After Foliowright: 624, and zero structural defects (tag tree, reading order, tables, lists, language, metadata) with zero PAC warnings.
Checked in PAC 2024, the tool accessibility offices actually open, and veraPDF, the ISO reference validator. Measured 7–9 September 2026. Method and results →
The 624 that remain are 331 images that need a person to write one sentence each, and 293 fonts the source never embedded — neither can be fixed by re-tagging. So PAC still shows a red banner, and Foliowright tells you exactly which pages the remaining sentences are on rather than inventing them.
Describe a chart, guess what a form field is called, or claim to fix a font the source never embedded. A fluent wrong description that still validates is the worst failure in this field, because nobody catches it. How it works →
Under the ADA Title II web accessibility rule, public universities and agencies must make their web content accessible, documents included: by April 2027 for entities serving 50,000 or more people, April 2028 for the rest. Most have years of PDFs in the backlog.
Desktop and self-hosted: your documents are processed on your own machine and, in the default mode, nothing is sent anywhere. See the Foliowright privacy policy.
Not a customer — there is nothing to buy yet. A design partner willing to share 20–50 real PDFs from its backlog and an hour of its time. hello@foliowright.com