Blind and low-vision shoppers buy clothes online too, and return them more, because nothing on a product page tells them what your garment looks like on them or whether their size will fit. Dress Aloud does both, out loud: it puts the item on the shopper's own photo and gives a plain, honest fit verdict. On your product pages it is one script tag. Behind your own experience it is an API.
Start a pilot: email mike@dressaloud.com
Pilot accounts are free for 60 days. Pricing after the pilot is per verdict; we will quote it in writing before anything is billed.
One of four verdicts, then the reasons. "Buy it in size L." "Buy a different size: M." "A word of caution before you buy size L:" followed by what could go wrong. "Do not buy this." only when the numbers say it will not fit in any size you offer. Then whether the size fits this body against your size chart, with the arithmetic quoted, what will go wrong where, and one practical alternative. Before the buy button: a return-risk level, low, medium or high, with the reasons, worked out in code from your chart, your fabric facts and the shopper's measurements.
A language model asked to judge fit will state measurements it was never given: a "typical" chest for a size, an invented sleeve length. A sighted shopper can notice; a blind shopper cannot. Dress Aloud runs every sentence of every verdict through a guard written in ordinary code: the only numbers allowed are your size chart's, the shopper's own measurements, and the differences between them. A sentence with any other number is sent back once, then removed, and the shopper is told a sentence was left out and why. We publish our measured results, misses included, at dressaloud.com/honesty. Your monthly record carries the same counts for your own traffic.
An accessible button, "Hear how this fits me, with Dress Aloud", that opens Dress Aloud with your item's facts already loaded from your catalog. The shopper's photo and measurements stay in their own Dress Aloud session; they never touch your page. Your page can listen for the verdict form and the return-risk level, nothing else. Your account pays for the try-on, so the shopper needs no plan of their own.
<script src="https://dressaloud.com/widget.js"
data-key="da_pub_YOURKEY" data-sku="TEE-001" data-size="L" async></script>You register each item's facts once (title, fabric, fit words, sizes, size chart, picture) with the API below, or by a nightly feed. Change the size with DressAloud.setSize("M") when the shopper picks one.
Server to server. You send the item's facts (or a registered SKU), the shopper's measurements and the size considered; you get back the verdict form, the guarded assessment text, the return risk with reasons, the size chart as spoken sentences, and the record of what the guard did. Verdicts for the same item, size and measurements are cached for six hours and not charged twice.
POST https://dressaloud.com/api/v1/verdict
Authorization: Bearer da_live_YOURSECRET
{ "sku": "TEE-001", "size": "L",
"shopper": { "height_in": 70, "weight_lb": 180, "chest": 43, "waist": 37, "prefer_fit": "regular" } }{ "verdict_form": "buy",
"assessment": "Buy it in size L. Your chest of 43 inches sits inside the chart's 42 to 44 inch range ...",
"return_risk": { "level": "low", "reasons": ["Your measurements sit inside the chart's range for size L."] },
"chart_sentences": ["Size M: chest 38 to 40 inches, waist 32 to 34 inches.", "..."],
"guard": { "flagged": [], "retried": false, "removed": 0 }, "remaining": 4999 }Your brand, your app or site, our engine. Per-tenant credit pool, rate limits, allowed origins and monthly records. Fit memory that follows your customer across devices can be held under your account with the shopper's consent, exportable and deletable on request. Ask for the integration guide.
Retailers buy accessibility to answer questions from counsel: was the accessible path there, was it used, did it work. Each month your account produces a record with counts: verdicts produced, verdict forms, how often the guard caught an unsupported number and how many sentences were removed, reported outcomes (fit as said, too big, too small, returned), failures and response times, and a statement of what data reached us. It is a paragraph of plain sentences plus the numbers, made to be filed.
| Through the API | Item facts and the shopper's measurements. No name, email, address, order or photograph. |
|---|---|
| Through the widget | Your item's facts, from your catalog. The shopper's photograph and generated pictures live in the shopper's own Dress Aloud session on their device; they are not stored on our server and never sent to you. |
| Back to you | Per try-on, the verdict form and the return-risk level, if your page listens. Per month, the record above. |
| Model provider | Google Gemini, as a service provider under our key. Item facts and measurements are sent to produce the verdict; nothing is used to train. |
Measured on real store pages with the guard on, and published with denominators at dressaloud.com/honesty. A try-on with a picture takes about half a minute; an API verdict without a picture is a few seconds. We will give you measured numbers for your own catalog during the pilot, not these repeated.
Email mike@dressaloud.com with your store's address and roughly how many product pages you have. You get a pilot account (API key, widget key, allowed origins, a credit pool) the same day, the integration guide, and a person to talk to. Try the widget on a sample page: the widget demo.
Dress Aloud for shoppers · Honesty results · Privacy · Terms. Dress Aloud is a creation of Always On Top Entertainment LLC, Port St. Lucie, Florida.