How it works

What "accessible" actually means here.

Every document is remediated against WCAG 2.1 AA and PDF/UA-1, the technical standard the DOJ's ADA Title II rule points to for PDFs — validated by machine, checked by a person.

The pipeline

From "as published" to "compliant," in nine steps.

Ingest

Documents arrive via a dedicated email-in address or a watched agenda URL. Every file is deduplicated by hash so nothing gets processed twice.

Classify pages

Each page is classified — text page, scanned page, table-heavy, form, map or drawing, mixed — so it's handled with the right approach.

Add a text layer

Scanned pages get OCR text layers added. Born-digital pages that already have selectable text are left alone.

Infer structure

Reading order, table structure, heading levels, and lists are inferred from the document's layout and adjudicated for accuracy.

Draft alt text

Images get a drafted description from their content and surrounding context; maps and drawings get longer "described-by" descriptions. Every piece of alt text is flagged for human review — none of it ships automatically.

Write the tags

The structure tree, reading order, table structure, alt text, artifacts, document title, and language are written into the file — along with bookmarks generated from agenda items.

Validate by machine

Every document runs through a PDF/UA rule-check profile. Hard-fail rules block delivery until they're resolved.

Human QA

A person reviews every piece of alt text, any low-confidence reading order, and a sample of pages — page image beside the tag tree, checked by hand before anything ships.

Deliver, and log it

You get the compliant file back. We append an immutable ledger entry: before/after file hashes, checks run, validator output, timestamps, and QA sign-off.

A deliberate choice

We repair the file you already published. We don't replace it.

Public records and exhibits have to look exactly as they were published — that's not negotiable for a legal record. So our default is repair-in-place: we add and correct tags, text layers, and alt text inside your existing file rather than regenerating it from scratch. The visual result is identical to what you published; what changes is what's underneath it.

Known hard cases we're built for

Scanned, low-quality legacy documents
Engineering drawings, site plans & zoning maps
Large multi-page financial tables
Multi-column newsletters
Handwriting & signatures (marked as image with description)
Redacted documents — we never OCR-reveal redacted content, and we'll flag improper redactions to you
Fillable forms, with every field labeled
1,000+ page packets, processed in chunks
Mixed-language sections

FAQ

Questions we get from clerks and IT directors.

No. Publish exactly as you do today — on whatever agenda platform or upload process you already use. We pick up new documents automatically and hand back the compliant version.

PDFs: agenda packets, minutes, forms, and reports. We focus on documents, not web pages — we don't offer HTML accessibility scanning or overlay widgets.

24–48 hours per document under our standard SLA, from ingestion to delivery.

Scanned and legacy documents get an OCR text layer plus the same tagging and human QA as everything else. Low-quality scans are a known hard case we design for.

WCAG 2.1 AA and PDF/UA-1 — the technical standards referenced by the DOJ's ADA Title II rule. Every document is machine-validated against a PDF/UA rule profile and spot-checked by a human with a screen reader.

We only accept public documents — published or intended for publication. Storage is US-region, we never train on customer documents, access is logged, and we'll delete your data on request.

See it work on a real packet.

Send us one of your recent agenda packets and we'll remediate it free of charge.

Get a Free Packet Review