Council minutes and agendas
Scanned on the office copier
No text layer, so a screen reader reads a blank pageYour website passed the audit. The 4,000 PDFs behind it did not.
Adaely reads every document linked from your site, tells you which ones a screen reader cannot open, fixes what a machine can fix, and sends the rest to a person.
Built for the clerks, IT directors, and communications leads of city, county, and state government, working toward the April 2027 and 2028 dates.
As posted todayAfter Adaely
City of Sample
Regular Meeting of Council, Minutes
| Ordinance | Motion | Vote |
|---|---|---|
| 2026-14 | Adopt | 5 to 0 |
| 2026-15 | Table | 4 to 1 |
zoning overlay, scanned image
6 blocking issues
- ×No structure tags0 tags found in a 14 page document
- ×Scanned image onlyNo text layer, nothing for a screen reader to read
- ×Reading order undefinedContent order falls back to the draw order
- ×Table has no header rowVote columns read as loose numbers
- ×3 images without alt textIncluding the zoning overlay
- ×Document language not setScreen readers guess at pronunciation
214days until April 26, 2027, for populations of 50,000 or more
580days until April 26, 2028, for populations under 50,000 and special districts
The DOJ moved the dates once and said it fully expects to enforce at the new ones. The standard did not move: WCAG 2.1 Level AA.
What is actually in a municipal document library
Nobody posts an inaccessible document on purpose
It happens because the copier scans to image, because Excel exports without tags, because the packet was merged from ten sources at 4pm. Here is what we find on almost every site we scan.
Annual budget and CAFR
Exported straight out of Excel, 180 pages
No tags, no headings, tables with no header rowZoning and land use maps
Saved as a picture
No alt text and no text alternative anywhere on the pagePermit and license forms
Flattened before posting
Fields unlabeled and impossible to complete by keyboardBoard and commission packets
Merged from several departments
Reading order scrambled, page numbering brokenPublic notices and bid documents
Whatever the vendor sent
Contrast below 4.5 to 1, headings faked with bold textThe work, in order
Five steps, and you only do two of them
Point us at your website
We crawl every linked document, including the ones buried three clicks deep in the 2019 archive. No plugin, no IT ticket, no access to your CMS.
You get an inventory
Every file, its failure list, its page count, and what remediation will cost. The inventory is yours to keep whether you buy anything or not.
We remediate
OCR for scans, structure tags, reading order, heading levels, table headers, form field labels, language, bookmarks, and drafted alt text.
A person reviews
Alt text and reading order get read by a human before a file ships. A model can describe an image. It cannot tell you whether the description is true.
You publish and keep the record
Remediated files come back with a conformance report per document, so when somebody asks what you did, you have an answer with a date on it.
How we handle your files
Public record still deserves careful handling
Where documents live
Microsoft Azure, United States regions only. Named regions on request for a procurement questionnaire.
No model training
Your documents are never used to train or fine tune a model, ours or anyone else's.
Retention you set
Pick a window. Deletion on request, confirmed in writing, including from backups on the stated schedule.
Standards we work to
WCAG 2.1 Level AA for web content and PDF/UA-1 (ISO 14289) for documents.
Procurement
W-9, certificate of insurance, and a plain language security summary ready to attach. Ask about cooperative purchasing and sole source paths.
Records requests
Remediated files keep the original text intact. We change structure, not content.
Things other vendors will not say out loud
Three honest limits
An overlay widget on your homepage does not fix a PDF. It cannot reach inside the file. If a vendor sold you one and called the project finished, your library is still where it was.
Automation handles structure well and judgment poorly. Tags, reading order, and table headers are mechanical. Whether alt text is accurate is not, which is why a person signs off.
No tool makes you immune from a complaint. What holds up is a documented, dated record of steady work. That is the thing we actually sell you.
From a pilot
We had 4,100 documents on the site and no idea which ones were a problem. The inventory alone was worth the money.
Clerk of Council, city of 21,000, Midwest
Start with the inventory
Know the work before you price the work
The initial scan costs nothing. You receive the document inventory, failure list, and page count. Remediation is scoped from what is actually in your library—no obligation and no surprise platform fee.
First stepFree
Complete website document scan and a thirty-minute review call.
Find out what is in your library. It takes a week and costs nothing.
Send us your website address. You get a full document inventory with a failure list and a page count, and a thirty minute call to walk through it. No obligation, and the file is yours either way.