Home › Blog › Why We're Building DeepTag
Build in PublicWhy We're Building DeepTag: A PDF Remediation Workspace
We're building DeepTag — and it's not what most "PDF accessibility software" turns out to be. It isn't an auto-tagger that promises to fix documents for you, and it isn't another checker that scores a file and walks away. It's a workspace for the people who actually do the remediation.
The problem: remediation is death by a thousand context switches
If you've ever remediated a real PDF by hand, this will be familiar. The work isn't hard because the decisions are hard — it's hard because the workflow is fragmented across tools that don't talk to each other:
- You jump constantly between Acrobat's Tags panel, a checker, and a screen reader.
- Table header scope gets set cell by cell, by hand, one attribute at a time.
- Reading-order fixes happen blind: drag an element, re-check, drag again, repeat.
- Every QA pass means restarting the screen reader to hear what changed.
- Every tool is built either for detailed remediation or for flagging problems — never both.
None of that is the skilled part of the job. It's friction around the skilled part — and it's where the hours quietly disappear.
The idea: see the fix and hear the fix in one place
DeepTag pulls the three things a remediator constantly cross-references — the tag tree, the table structure, and the screen-reader output — into a single view. The goal is simple to state and surprisingly rare in practice:
When the tag tree, the table you're editing, and what a screen reader actually announces are all in front of you at once, the loop collapses. No window-juggling, no blind dragging, no restarting JAWS to check a single change.
What it does not do
DeepTag does not make the accessibility decisions for you. Whether a graphic is a meaningful figure or decoration, what scope a table header should have, how the reading order should actually flow — those are human judgement calls, and they stay that way. That's the same principle behind our remediation service: a document nobody reviewed is a document nobody should certify. DeepTag removes the busywork so that judgement gets more of your attention, not less.
Where it's at
DeepTag is in active development, built and tested inside our own remediation practice on real client work to PDF/UA, HSS and WCAG 2.0/2.1/2.2. If you remediate PDFs for a living and want to follow along — or try an early version — we'd genuinely like to hear from you.
Interested in early access to DeepTag?
Tell us a little about your remediation workflow and we'll be in touch.
Get in touch →