Foliowright

PDF accessibility remediation, for institutions that have to prove it

Design partners

We are looking for one accessibility office

Not a customer — there is nothing to buy. A design partner: one office willing to put real files in front of this and tell us where it is wrong. Everything we have measured was measured against public documents we chose ourselves, which is the weakest kind of evidence there is.

Who this is for

A public university or agency under ADA Title II or Section 508
With a real remediation backlog and a deadline attached to it. April 2027 for entities serving 50,000 or more people; April 2028 for the rest.
An office that already runs PAC or Acrobat's checker
Because then we are measured against the benchmark you already trust, not one we chose. If you have existing reports on your files, those are the most useful thing you can send.
Someone who has been disappointed by auto-tagging before
You are the reader this was built for. You know what a tool that produces confident wrong output costs, because you have had to clean up after one.

You do not need to switch tools, sign anything, or install a service. This is a conversation and a few files.

What we ask for

  1. 20–50 PDFs from your backlog

    Syllabi, policies, board minutes, forms, financial tables, slide exports. The messy real ones, not your best examples. De-identified is fine.

  2. Whatever reports you already have on them

    PAC, Acrobat, CommonLook. This is how we get measured against your benchmark rather than ours.

  3. One hour showing us how you remediate a document today

    Timed. The number that decides whether this product is worth anything is minutes per document, and we do not have yours — only our own guesses, which is not the same thing.

  4. Thirty minutes, later, reacting to our output on your files

    Is it correct? Is the review faster than what you do now? Where does it get things wrong? The last question is the most useful one.

What you get, and what we will not do

Your files, remediated, with the residual named
Not just output — a per-document account of what is left, why, and which page it is on. Including the parts we could not fix.
Your priorities shape what gets built
The roadmap is currently ordered by our judgement about documents we chose. That is a bad way to build a tool for your workflow.
We will not name you without asking
No logo, no case study, no quote, no “trusted by” — nowhere, until you have said yes in writing. This page will say when there are partners; it will not say who unless they want it to.
Your documents are not training data
They are used to test the engine against real material and deleted on request. No model is trained on them.
No compliance guarantee, from us or anyone
PDF/UA and WCAG depend on judgements a tool cannot make. Using this does not by itself make a document conformant and does not discharge your legal obligation. Nobody's tool can. The terms say so too.

Where this honestly stands

The engine is validated and the product is not proven. Those are different claims and it matters which one you are being told.

What is true: on six of the hardest real documents we could find, PAC 2024 reports no machine-fixable defect and no warning, and everything left is a figure awaiting a sentence or a font the source never embedded. The runs are published.

What is not true yet: that this saves anyone time. No accessibility office has used it. We have never watched a coordinator work through its review queue. The number we most want — minutes per document, before and after — does not exist, and cannot until someone lets us measure it.

That is the whole ask. One office, a few files, an hour.

Start a conversation