In February 2026, Sea Limited struck a partnership with Google on “agentic shopping prototypes.” By June, Shopee had become the first Southeast Asian e-commerce platform built directly into ChatGPT, live across Indonesia, Malaysia, the Philippines, Singapore, Thailand, Taiwan and Vietnam. Lazada, meanwhile, built its own AI shopping agent, Lazzie, running on Alibaba’s Qwen model, and credits it with a measurable lift in engagement during flagship sales events.

Two of the region’s biggest commerce platforms have quietly redefined who, or what, the customer actually is.

The customer might not be a person anymore

For two decades, selling online in Southeast Asia meant winning a person’s attention: a good photo, a fair price, a five-star review a shopper could read for themselves. Increasingly, it means winning an algorithm’s judgment instead. When a shopper asks ChatGPT to find “a good baby stroller under SG$150 (US$117),” it is the agent that reads the listings, compares the options and applies whatever criteria it was built with. Only the winner it picks ever reaches the human.

Consumers seem ready for exactly this handoff, in precisely the categories Southeast Asia’s small sellers depend on most. A June 2026 Checkout.com study of consumer and merchant readiness found shoppers are most comfortable delegating grocery shopping (41 per cent) and household supplies (31 per cent) to an AI agent, the bread-and-butter categories for millions of small Shopee and Lazada sellers, while trust drops sharply for higher-stakes categories like financial services (15 per cent). The same study found 89 per cent of merchants are already preparing for agentic commerce, even though only 3 per cent of transactions today actually involve an AI agent. Readiness is running well ahead of reality, which is usually when the ground shifts fastest.

Also Read: Southeast Asian tech leaders are learning to trust AI agents, but not with production

Selling to an algorithm nobody can audit

Here is the uncomfortable part for the region’s small sellers: nobody outside Shopee, Lazada, OpenAI and Alibaba can see how these agents actually decide what to surface. A merchant can optimise a listing for a human eye, with better photos, clearer copy, faster replies, because a human’s attention is at least partly legible. An AI agent’s ranking logic is not published, is not stable, and is not something a small seller has any real way to contest if it quietly stops recommending their shop.

The market is not waiting for sellers to catch up. The global AI shopping assistant category is projected to grow from roughly US$4.2 billion in 2025 to US$22.1 billion by 2032, and in Southeast Asia, AI recommendations are already estimated to influence 90 per cent of purchase decisions, with 80 per cent of shoppers engaging these tools weekly. That is not a future scenario. It is the market Southeast Asia’s sellers are already operating inside.

To be fair, this is not a story of platforms acting in bad faith. Shopee’s OpenAI partnership includes ChatGPT for Business trials for eligible sellers, meant to help them write listings, marketing copy and automate customer service. That help matters, but it addresses the seller’s side of the conversation, not the agent’s decision logic on the other end of it.

Even consumers, who stand to gain the most convenience, are not handing over full control: Checkout.com’s respondents wanted spending caps, instant revocation and easy cancellation before they would trust an agent to buy on their behalf, and on average would only let one spend around £177 (US$237) without asking first. If shoppers are this cautious about what an agent can do with their own money, sellers are entitled to at least the same caution about what an agent can do to their livelihood.

Also Read: The hidden economics of autonomous AI agents

The capital isn’t going where the sellers need it

It would help if the region’s AI investment boom were flowing toward the unglamorous work of making these systems legible to the small businesses that now depend on them. Mostly, it isn’t. Southeast Asia’s native AI startups raised a record US$4.1 billion in the first seven months of 2026, more than double all of 2025, but roughly 68 per cent of that sits in a single transaction, Kling AI’s US$2.8 billion round for its generative-video platform. Strip that one deal out and regional AI funding is closer to US$1.3 billion, concentrated in infrastructure and foundation models, not in tools that would let a small seller understand why an agent stopped recommending their stall.

I did not train as an engineer. I trained as a paramedic, then taught myself to build software because I wanted ordinary people, not only institutions, to have a fair shot at seeing what’s coming before it arrives, an instinct that eventually carried me from self-taught builder to founder and, later, to a board and product-architecture role at a Singapore culture-tech venture. That background is why this particular shift concerns me more than most AI headlines do. The risk was never that AI agents would simply replace shoppers. It’s that they already have, in exactly the categories where the region’s smallest sellers make their living, and almost nobody has given those sellers a way to see, question, or appeal what the agent decided.

Southeast Asia spent a decade teaching hundreds of thousands of small sellers how to win over a customer. It now owes them a fair shot at winning over the algorithm that has quietly taken that customer’s place.

—

Editor’s note: e27 aims to foster thought leadership by publishing views from the community. You can also share your perspective by submitting an article, video, podcast, or infographic.

The views expressed in this article are those of the author and do not necessarily reflect the official policy or position of e27.

Join us on WhatsApp, Instagram, Facebook, X, and LinkedIn to stay connected.

The post Southeast Asia’s sellers spent a decade learning to sell to people. Now they have to sell to AI appeared first on e27.