Agentic commerce means AI agents that research, compare and complete purchases on a shopper’s behalf. Alipay’s AI Pay processed more than 120 million of those transactions in a single week of February 2026, analysts size the global opportunity at US$3 to US$5 trillion by 2030, and 51 per cent of Canadian shoppers say they would let AI finish the sale.
Between February 5 and 11 of this year, Ant Group’s Alipay AI Pay service cleared more than 120 million transactions in seven days. Every one was completed by an AI agent rather than a person tapping a screen. Measured against Canada’s population that is roughly three purchases each, inside a single unremarkable week. NielsenIQ has built its new global report, The Commerce Revolution: Where East Meets West, around that figure. For anyone selling jewellery, the implication is narrower than the headline: software is doing the buying now, and it can only buy what it can find.
What is agentic commerce and why does it matter now?
Agentic commerce describes AI systems that move past recommending products and begin acting on them: researching options, comparing prices, filling carts and, with permission, paying. NielsenIQ’s report treats it as the layer now joining live commerce, social commerce and quick commerce together, and the sums attached to that view are large. McKinsey sizes the global opportunity at US$3 to US$5 trillion by 2030. Morgan Stanley projects agents will drive US$190 to US$385 billion of American e-commerce by the same year, roughly 10 to 20 per cent of the total.
The demand side has moved faster than most forecasts allowed. Alibaba’s Qwen assistant passed 100 million monthly active users two months after launch. Adobe Analytics recorded a 1,200 per cent rise in traffic reaching American retail sites from generative AI sources between July 2024 and February 2025. Those visitors went deeper into the site and bounced less often than traffic from any traditional channel. They still convert at a lower rate, although Adobe’s later data shows the gap closing each quarter.
Emilie Darolles, NielsenIQ’s president for Western Europe, describes a gap between the two behaviours: shoppers are increasingly comfortable letting AI research and recommend, and still reluctant to let it pay. Whoever earns enough trust to be picked by the shopper and by the agent, she argues, takes the next phase.
How close are Canadian shoppers to handing AI the sale?
Closer than most of the trade assumes. Adyen’s 2026 Retail Report, released in January, found 51 per cent of Canadian consumers would allow AI to manage the entire shopping process, checkout included. Thirty per cent already use AI shopping tools, nearly triple the share from a year earlier, with millennials leading at 56 per cent willingness and boomers trailing at 35. A National Bank survey conducted by Léger in June 2026 adds the influence layer: 39 per cent of Canadians used generative AI to support a purchase decision in the past year, and among those users, 61 per cent say AI now sways their buying more than advertising does.
Globally, NielsenIQ’s own consumer data keeps the picture honest. Around 34 per cent of consumers use AI to research products and 23 per cent to summarise reviews, yet only about 8 per cent have let an agent complete a purchase. Influence is running far ahead of automation. No Canadian data yet isolates jewellery within these numbers, and readers should know that gap exists; the figures above cover retail broadly.
| Measure | Figure | Source |
|---|---|---|
| Alipay AI Pay transactions, week of Feb 5 to 11, 2026 | 120 million+ | Ant Group |
| Global agentic commerce opportunity by 2030 | US$3 to $5 trillion | McKinsey |
| Share of US e-commerce driven by agents by 2030 | 10 to 20 per cent | Morgan Stanley |
| Growth in generative AI traffic to US retail sites, Jul 2024 to Feb 2025 | 1,200 per cent | Adobe Analytics |
| Canadians open to AI completing checkout | 51 per cent | Adyen 2026 Retail Report |
| Canadians who used generative AI to guide a purchase in the past year | 39 per cent | National Bank / Léger, June 2026 |
| Of those users, share who say AI sways buying more than advertising | 61 per cent | National Bank / Léger, June 2026 |
| Consumers globally who have let AI complete a purchase | ~8 per cent | NielsenIQ |
Why does jewellery feel this shift differently?
An engagement ring behaves nothing like a weekly grocery order. It is bought once or twice in a lifetime, after long consideration, and it carries feeling that no comparison engine handles well. That is the reason the 8 per cent autonomy figure offers less comfort than it appears to. The exposure for jewellers sits earlier, at discovery. Once an agent has assembled a shortlist of three rings, the old sequence of impression, click and cart collapses into one decision made by machine. A store whose product pages carry no structured data, whose prices say “call for details” and whose certificates sit inside scanned PDFs will not appear on that shortlist, for the plain reason that an agent cannot shortlist what it cannot read.
The rails being laid for this future have a Canadian signature. Ottawa-based Shopify, which powers thousands of Canadian jewellery storefronts, has spent the past year pushing its open Universal Commerce Protocol and surfacing its merchants inside ChatGPT, while OpenAI reshuffled its Instant Checkout plans in March toward branded apps. Shopify president Harley Finkelstein’s argument for why merchants win rests on data depth from millions of stores and billions of transactions. Independent jewellers on those platforms will inherit agent readiness sooner than most of them realise. The money side of the same shift is already covered in Canadian Jeweller’s report on agentic AI in cross-border payments, where the agents run the wires instead of the carts.
What should jewellers do before the agents arrive?
The Monday-morning file has five entries. First, audit product data: every piece online needs structured fields for price, metal, carat weight, certification and availability, marked up in schema.org Product format so agents can parse it. Second, run the mirror test. Ask ChatGPT, Perplexity and Gemini where to buy an engagement ring in your city, and read what they say, and do not say, about your store. Third, move essential facts out of images and PDFs into crawlable text, certificates and return policies first. Fourth, anyone selling on Shopify should watch its agentic checkout betas, because that pipeline will carry Canadian independents into AI storefronts ahead of custom-built sites. Fifth, protect the human close. The data shows shoppers still want a person at the moment of yes, so train staff for the customer who walks in holding an AI-generated shortlist, because that customer has already done the comparison work and arrives ready to buy or to leave.
China has supplied the proof that the demand exists. The West is still building the trust rails underneath it. Canadian jewellers control neither, and they do control the one thing that decides whether they are picked: whether their inventory can be read by the systems now making the choice. February was the first week that showed up on the record. It will not be the last.
Frequently asked questions
What is agentic commerce?
It is commerce carried out by AI agents acting for a shopper: researching products, comparing options, filling carts and completing payment within limits the shopper sets. It goes beyond chatbots that only recommend.
How big is agentic commerce expected to become?
McKinsey sizes the global opportunity at US$3 to US$5 trillion by 2030. Morgan Stanley projects agents will drive 10 to 20 per cent of American e-commerce by the same year.
Are Canadian shoppers ready to let AI buy for them?
Adyen’s 2026 Retail Report found 51 per cent of Canadians would let AI manage the full shopping process, including checkout, and 30 per cent already use AI shopping tools. Actual autonomous purchasing remains rare, at roughly 8 per cent globally.
Why does this matter for jewellery stores specifically?
AI agents shortlist products from structured data. Jewellery sites that hide prices, rely on images and bury certificates in PDFs cannot be read by agents, so they never make the shortlist a customer sees.
What is the first step a jeweller should take?
Add schema.org Product markup with price, metal, carat and certification to every online listing, then ask AI assistants about jewellery stores in your area to see how your business currently appears.
Sources: NielsenIQ, The Commerce Revolution: Where East Meets West; Ant Group via Business Wire; Adobe Analytics; Adyen 2026 Retail Report; National Bank / Léger survey via Retail Insider; Digital Commerce 360.
Author: Olivier Felicio
Olivier Felicio is the founder and publisher of Canadian Jeweller Magazine, Canada's jewellery business publication since 1879. He writes on retail strategy, market data, trade policy and the commercial side of the Canadian jewellery and watch trade, and hosts Time and Shine, Canada's jewellery buying event. Contact: olivier@canadianjeweller.com.
