Accessibility
Accessibility statement
We build to WCAG 2.2 Level AA, we test it on every change, and here is exactly how far that has been proved.
Version 1.2 · In effect from 27 October 2026 · Download as PDF
What we hold ourselves to
The Web Content Accessibility Guidelines 2.2 at Level AA, across everything we publish: this site, the portal a school’s staff sign in to, the pages a caller reaches from a text message, and the tools our own staff use. Many of our customers take federal money and carry Section 504 and 508 obligations that reach their vendors. We intend to meet them, and we would rather be checked than trusted.
What is in place
Everything works from a keyboard, and whatever has focus shows a visible ring. Dialogs keep focus inside them while they are open and hand it back when they close.
Text meets the contrast WCAG requires, measured rather than judged by eye. Body text and every small label clear 4.5 to 1 against the background they sit on. The one lighter grey is kept to text 24 pixels and larger, where the requirement is 3 to 1, and it clears that on every background it is used on. There is one theme, and it is light: it stays light when your device asks for dark mode, so what we measured is what you see.
Every form field has a real label, errors are announced and tied to the field they belong to, and status is never shown by colour alone.
Motion is optional. If your device asks for reduced motion, nothing on this site animates. This site is also complete with JavaScript switched off.
Every recording has a transcript. We are a voice company, and a recording nobody can read is a recording half our users cannot use.
How we test it
An automated accessibility scan, axe-core, runs over every page of this site, every screen of the portal and every staff tool but the two named below. It is part of the test suite that runs on every change proposed to the code, and a violation fails that suite. Each screen is scanned twice, with the device set to light and then to dark, to prove that nothing changes underneath somebody who prefers dark mode. The last full run was on 21 September 2026. A second suite walks the core tasks with no mouse at all — including asking for a demo on this site — because a scan can prove a control is labelled and cannot prove that a person can reach it.
This site is checked at widths from 320 pixels up, so nothing scrolls sideways on a phone, and every button and link on it is at least 44 pixels tall.
What we have not done yet
Nobody who uses a screen reader daily has tested the product. Automated tools find roughly a third of accessibility problems; the rest are judgements about whether a screen makes sense, and ours so far rest on design review. This is the largest gap and we say so.
A reschedule link expires without warning. The link a caller receives by text stops working after a set time, and the page gives no notice beforehand.
The portal is not yet tested at 320 pixels wide. This public site is; the signed-in screens are held to the same rule by design review rather than by a test.
Tooltips and sticky bars are not yet under test. We believe every tooltip can be dismissed with Escape and that no sticky element covers a focused control, and neither belief is proved by a machine yet.
Two staff screens are not scanned yet: a single onboarding run and a single prospect. Both are internal tools our own staff use; no school and no caller sees them.
The conformance report
A full Accessibility Conformance Report in the VPAT 2.5 format — every WCAG 2.2 Level AA criterion, what supports the claim for it, and the known limitations gathered in one place — is maintained alongside this statement and is available on request from the address below. It is written from test results rather than from intent, and it is the document a procurement office should ask for.
Tell us
If something here does not work for you, write to support@doras.ai. Say what you were trying to do and what happened; if it is a defect we will fix it, and we will tell you when we have.