For
PDF remediation tooling for accessibility consultants
You bill for judgement, not for dragging tags. What a tool should do for you, and what it should refuse to do on your behalf.
What you are actually selling
Not remediation. Judgement, and a defensible record of it.
Any competent tool can find a missing /Alt. What a client is paying you for is the decision that this chart is informative and that flourish is decorative, that this grid is a data table and that one is layout, and that this document is now in a state you are willing to put your name to.
A tool that tries to make those decisions for you is not saving you time — it is creating work, because you have to check every one it made.
What a tool should refuse to do
Taggart’s design has a specific line, and it is worth stating because it is unusual.
It will not write alt text without your approval. Every generated description arrives as a draft in the review queue, and the decision is recorded.
It will not artifact content on a heuristic. Artifacting removes content from the reading order entirely, and a tool that does that on a guess will eventually delete a heading. Candidates are listed; you approve them.
It will not write a PDF/UA conformance claim into a file its own checker cannot confirm. Ask for it with blockers outstanding and the report says why it declined.
It will not invent a ToUnicode mapping it cannot recover. A wrong one silently corrupts every extraction from that point and looks conformant while doing it.
What it should do fast
Find everything mechanically, mapped to the Matterhorn condition, the PDF/UA clause and the WCAG criterion, so your report writes itself.
Fix the mechanical majority in one pass, with a diff and one undo.
Make reading order fast. Numbered badges on the page, keyboard reordering, batch undo, and a screen-reader preview beside it.
Prove it did not damage anything. Every output re-rendered and compared to the original at SSIM ≥ 0.999.
Produce the record. Every human decision with an actor and a timestamp, the tool version, and the SHA-256 of both files.
The Hebrew case
If you work with Israeli documents, the visual-order problem is the largest single item in your queue and no other tool sees it. A Hebrew PDF stored backwards passes PAC, passes veraPDF, passes Acrobat’s checker, and reads as nonsense.
What the failure is and how it is repaired →
Frequently asked questions
Will a tool like this undercut what I charge?
It changes what you charge for. Clients are not paying for tag dragging; they are paying for someone who can look at a table and know whether it is data or layout, and stand behind the answer. Doing the mechanical part in eight minutes instead of forty means more documents per day at the same rate, not a lower rate.
Can I brand the remediation report?
Custom report branding is in the Pro tier. The report is also available as JSON, so if you have your own deliverable format it is a transform rather than a re-entry.
I work in Hebrew. Is this actually built for that?
Hebrew is a first-class locale, not a translation. The interface is Hebrew with a right-to-left layout, the editor knows a two-column Hebrew page reads right column first, the engine detects and repairs visual-order storage, and IS 5568 is a selectable profile. It is the reason the product exists.
Does it run on macOS?
It runs in a browser, so yes — which is more than can be said for axesPDF, CommonLook or PAC.
Check your own file. Taggart's validator is free and unlimited — no page cap, no watermark, no account.
Last updated .