1. What actually launched
Hey Savi is an AI-powered fashion search platform built on computer vision and conversational AI. A shopper feeds it a screenshot, a photo, or a text description, and it returns shoppable matches across more than 10,000 brands — ranked, according to the companies, on relevance rather than sponsored placement. PayPal's "Agentic Commerce Services" supply the payment and merchant-connection layer underneath it, making product data — pricing, images, inventory — accessible to the platform and enabling checkout without the shopper leaving the Hey Savi app.
Debenhams Group is the first retailer to plug in, across four of its owned brands. As Mike Edmonds, PayPal's Vice President of Agentic Commerce, put it: "Shopping now starts with a screenshot or a creator post, but the path to purchase doesn't move at the same speed." The product exists to close that gap.
2. Why this is a genuine milestone, not another AI-shopping press release
Fashion has seen plenty of "AI shopping assistant" announcements that amount to a chatbot bolted onto a search bar. This is different in one specific way: it is described as the UK's first native in-app checkout for agentic fashion commerce, with real, recognisable high-street brands live on day one, not a pilot with an unnamed retailer. It follows a broader pattern flagged in the McKinsey/Business of Fashion State of Fashion 2026 report — see our industry trends analysis — that agentic commerce is arriving faster than most retailers' product-data infrastructure is ready for.
3. The problem nobody's launch copy mentions: fit
Every agentic-commerce pitch is framed around speed: fewer steps between seeing something and owning it. For most product categories that's a clean win. Fashion is the exception, because the single biggest cause of hesitation and returns — will this fit, will it suit me, is this the right size in this specific cut — is not solved by removing checkout steps. If anything, compressing "screenshot → owned" into one motion removes the moments that used to partially answer that question: reading a size guide, comparing product photos from multiple angles, checking reviews for fit notes.
Put plainly: agentic checkout without fit data is a faster way to generate a return. The conversion upside of agentic commerce and the return-rate downside of unresolved fit uncertainty are pulling in opposite directions, and right now most retailers' product feeds are built for the first and silent on the second.
4. What this means for fashion brands' product data
An agent completing a purchase on a shopper's behalf can only be as good as the data it can read. If sizing and fit information live only as prose in a size-chart tab, an agent parsing a product feed has nothing structured to reason with — it can match a screenshot to a product, but it cannot tell a 5'4" shopper that this coat runs long, or that this brand's "medium" fits closer to a competitor's "small." That gap doesn't disappear because checkout got faster; it just moves downstream into a return.
This is the same structural point the State of Fashion 2026 report makes about AI visibility generally: semantically rich, API-accessible data determines whether an agent can act on a brand's behalf with confidence. For fit specifically, that means moving beyond a size chart as a paragraph of text, toward fit and sizing signals a machine can actually parse — the same direction we've written up in detail in our agentic commerce PDP checklist.
5. What to watch next
- Whether Hey Savi/PayPal publish return-rate data from the Debenhams Group rollout — the first real test of whether agentic checkout changes fashion return rates up or down.
- Whether other UK or US retailers follow with their own agentic-checkout integrations before the fit-data question is addressed.
- Whether size/fit data becomes a standard field in agentic commerce product feeds, the way price and inventory already are.
6. True Fit's agentic fit agent: the other half of this story
Hey Savi/PayPal solved the checkout step. A separate, equally significant launch solved a different piece of the same puzzle: on 17 February 2026, True Fit announced an "agentic AI shopping experience" built on nearly 20 years of purchase and returns data — reportedly $616bn+ in analysed transactions across 91,000+ apparel and footwear brands — available to early-adopter retailers from 1 March 2026 with broader release in April. It exposes that data via Model Context Protocol (MCP), so external AI agents and shopping copilots can query structured fit and size guidance directly, and the agent is designed to intercept "will this fit?" hesitation in real time rather than let a shopper order three sizes to compare.
True Fit is not a new entrant reacting to the agentic-commerce wave — it has run a self-serve, no-code Shopify app since July 2023 (4.0★ from 19 reviews on the Shopify App Store, pricing from $1,000/month with volume tiers beyond that). The 2026 agentic layer is a capability added to existing retailer infrastructure and an existing data asset, not a new go-to-market.
Read alongside the Hey Savi/PayPal launch, the pattern is consistent: agentic commerce is moving fast on speed and discovery, and the fit-data layer is where the two leading approaches diverge. True Fit predicts fit from historical purchase and return data — a size recommendation with no visual component anywhere in its public product. Rendered Fits shows the garment photorealistically on the shopper's own body, and is installable self-serve on Shopify today, typically from £149/month — no enterprise contract, no $1,000/month floor. The two are answering genuinely different shopper doubts: "what size do I need" versus "how will this actually look on me." Rendered Fits also combines that visual layer with Complete-the-Look (multi-item outfit try-on) and AI size recommendation in one widget, live on production today — a combination that has no equivalent in True Fit's public feature set, which remains single-purpose size/fit prediction.
For brands weighing where to invest: statistical fit prediction and visual fit proof are complementary, not competing, layers of the same fit-confidence problem. A brand chasing agentic-commerce readiness plausibly needs both a structured data layer an agent can query and a visual layer that gives the shopper — human or agent-assisted — genuine confidence before checkout.
Frequently asked questions
What is the Hey Savi and PayPal agentic commerce launch?
On 2 June 2026, Hey Savi and PayPal launched what they describe as the UK's first agentic commerce platform with native in-app checkout. Hey Savi is an AI-powered fashion search platform covering 10,000+ brands; PayPal supplies the Agentic Commerce Services and payment layer. Debenhams Group — Debenhams, Karen Millen, Boohoo, and Pretty Little Thing — is the first retail adopter.
Does agentic commerce increase or decrease fashion return rates?
It depends on whether fit-confidence data is available to the agent at checkout. Faster, screenshot-to-purchase flows remove the manual steps — size charts, multi-angle photos, review reading — that normally reduce fit-based returns. Without structured fit data in the loop, faster agentic checkout risks higher returns, not lower.
What is agentic commerce in fashion retail?
AI-driven shopping where an agent takes a shopper from discovery to purchase — often from a screenshot or description rather than a search query — within a single flow, sometimes completing checkout on the shopper's behalf.
How can fashion brands prepare their product pages for AI shopping agents?
Ensure sizing, materials, and fit information exist in structured, machine-readable form (schema.org Product/Offer markup at minimum), not only as page copy. See our agentic commerce PDP checklist for implementation steps.
Is Rendered Fits the same as True Fit?
No. True Fit predicts size and fit from roughly 20 years of purchase and returns data, delivered as conversational or numeric guidance with no visual component. Rendered Fits shows the garment photorealistically rendered on the shopper's own body, self-serve on Shopify from around £149/month. They address different shopper doubts — "what size" versus "how will it look" — and can be used alongside each other.
Sources
- PayPal Newsroom — "Hey Savi and PayPal Launch UK's First Agentic Commerce Platform with In-App Checkout; Debenhams Group Joins as First Retail Adopter" (2 June 2026). Launch details, Mike Edmonds quote, Debenhams Group brand list. newsroom.paypal-corp.com
- McKinsey & Company / Business of Fashion — The State of Fashion 2026. Agentic AI and semantically-rich-data framing for AI-agent visibility. businessoffashion.com
- Businesswire — "True Fit Launches Agentic AI Shopping Experience Powered by 20 Years of Fit Data" (17 February 2026). Launch date, data-scale claims, MCP positioning. businesswire.com
- True Fit — Shopify App Store listing. App Store launch date, 4.0★/19 reviews, pricing from $1,000/month. apps.shopify.com/truefit
This page reflects publicly reported facts as of the launch dates and will be updated if return-rate or performance data is published. Last updated 3 July 2026.