Where Meta’s WhatsApp agent can actually win

In Hangzhou or Shenzhen, the answer is that your order shows up, often within the hour. China fused messaging, payments and shopping into single apps more than a decade ago. WeChat carries roughly 1.4 billion users, an in-app store layer with hundreds of millions of monthly shoppers, and a wallet most of the country pays with. Korea built its own version, where KakaoTalk made chat the default way to send a gift. This is the world Meta gestures at when it imagines what WhatsApp could be.

And yet WeChat, the purest “messaging app does commerce” story, is not actually China’s shopping champion, even though it arrived first and is still the bigger app. People do not open a messaging app to browse and shop. The buying went instead to Douyin, the Chinese app run by TikTok’s owner ByteDance, whose endless video feed is engineered to make you want things you were not looking for. WeChat had the users and the wallet, and it still lacked the two things that actually move commerce: A feed that creates demand and a way to deliver the goods. A chat window is neither.

Korea shows what a messenger can and cannot win. KakaoTalk is the country’s WhatsApp, and it owns one kind of commerce completely: gifting. Koreans send presents straight from the chat window, close to 200 million of them in 2025, which is nearly all of the country’s mobile gifting. But notice what kind of commerce that is. A gift voucher or a coffee coupon needs no warehouse. The moment a purchase becomes a physical thing that has to arrive fast, the winner is no longer the messenger but Coupang and its dawn-delivery network. KakaoTalk owns the commerce that fits inside a message; Coupang owns the commerce that needs a truck.

Japan makes the same point in the negative. LINE is about as dominant a messenger as exists anywhere, reaching 97 million people, close to 78 percent of the country. If messaging reach alone turned into commerce, LINE would own Japanese retail. Instead, it shut down its own payments service in 2025 and handed the wallet to a rival, while the actual shopping stayed with Rakuten and Amazon Japan. The most-used chat app in the country could not turn that reach into owning what people buy.

Every market tells the same story: A chat app does not win physical commerce. Whoever owns the warehouse does.

Which brings us back to the WhatsApp agent, where Meta starts further ahead than WeChat ever did. Through Instagram and Reels it owns the demand-making feed WeChat never had, the agent gives it the sales conversation, and in the West, paying by card is universal. Only the last pillar is missing. Meta has no warehouses, no trucks, no delivery promise of its own and the few times it reached for the pieces around the sale, it pulled back: Its own wallet, Meta Pay, never became something people use, and in 2025 it wound down in-app checkout for Facebook and Instagram Shops, sending buyers back to merchants’ own sites to pay, ship and handle returns. Even Marketplace, its billion-user listings surface, mostly stays out of the transaction itself.

And in the West, that last pillar is already spoken for. The West did fuse commerce, just not around chat. Amazon long ago combined the storefront, the payment, its own branded credit cards and the expensive part, the warehouses and the trucks, into one app that owns the American purchase from search to doorstep. That is the same kind of vertical integration China’s commerce giants built, with players like Alibaba and JD racing into a market where no Amazon yet stood in the way. In the US, that lane was filled years ago.

None of this makes the agent a mistake. It is already a booming ad business for Meta, and maybe that is all Meta wants it to be: Commerce’s front door, sending the shopper onward and billing the merchant for the introduction.

But goods are only half of commerce, and the other half never needed a warehouse. Remember what KakaoTalk won: Gifting, the one kind of buying that ships nothing. Services are the same, only far bigger. A haircut, a dental cleaning, a training session, a plumber’s visit, a tutor’s hour: None of it sits in a fulfillment center. The transaction is a booking, not a box.

And a booking is exactly what the agent is built to take. Look at the feature Meta put in its own announcement, right beside answering questions and recommending products: It books appointments. For a salon, a clinic or a one-person studio, that is a front desk. Give it the two pieces still missing, a calendar to hold the schedule and a way to take payment inside the chat, and WhatsApp stops being where those businesses message customers and becomes where they run the day.

None of it needs a warehouse, and none of it is Amazon’s to defend. Does that put Meta on a collision course with Square and Mindbody?