By 2030, AI shopping agents are projected to drive up to US$1 trillion of retail revenue in the United States alone, with global estimates running as high as US$3 to 5 trillion. McKinsey, Salesforce, and JP Morgan have published independent forecasts within the same range in the past six months.
The surprising part is not the size of the number. The surprising part is who decides where that money goes: not the customer, but the customer's AI. This guide explains what agentic commerce is, how it actually works, and what every Hong Kong retail and e-commerce owner needs to do this year to stay visible in the new buying flow.
What Is Agentic Commerce?
Agentic commerce is a buying model in which an AI agent acts as the customer's personal shopper. The agent receives a goal in natural language ("buy me running shoes for under HK$800"), researches options, compares prices and reviews across merchants, makes the purchase decision, and completes the checkout. The human approves preferences and budget, not individual transactions.
Think of it as the shift from "search and click" to "describe and approve". The customer never sees your website. The AI does.
How Does Agentic Commerce Actually Work?
An agentic commerce transaction follows the same five steps every time. Once you understand them, the implications for your store become obvious.
The 5-step transaction flow:
— Goal capture: the customer tells the AI what they want, often in conversational language with constraints (price, delivery time, brand preferences).
— Discovery: the AI queries multiple merchants, marketplaces, and product feeds. This is where your store either appears or does not.
— Evaluation: the AI compares options against the customer's stored preferences and the stated constraints, scoring each candidate.
— Authorisation: the AI either auto-purchases within a preset budget or asks the human to approve a shortlist. The trend in 2026 is towards larger pre-authorised budgets.
— Fulfilment: the AI completes checkout using stored payment, triggers delivery, and tracks the order on the customer's behalf.
Who Are the Major Players in 2026?
Five platforms now drive the bulk of agentic commerce traffic. Hong Kong retailers should understand each one and decide which to optimise for.
The 2026 player landscape:
— ChatGPT Instant Checkout: launched late 2025, expanded through 2026. Allows ChatGPT users to buy directly inside the chat without leaving for a merchant site. Now connected to Stripe and partner networks.
— Perplexity Comet Browser: an AI browser that researches, compares, and purchases on the user's behalf, launched in early 2026.
— Amazon Rufus: Amazon's in-app shopping agent that recommends and orchestrates purchases inside the Amazon ecosystem.
— Google AI Mode (Shopping): Google's agentic shopping experience, integrated with the Merchant Centre product feed.
— Shopify's Agent Storefront: Shopify announced in March 2026 that its merchants are being upgraded to support direct agent-to-store transactions.
Why Should a Hong Kong SME Care Right Now?
Three forces converge in 2026 that make this a real-world commercial issue, not a future-of-retail thought experiment. The Hong Kong retail and e-commerce market is small, exposed to cross-border traffic, and especially vulnerable to discovery-layer disruption.
Force 1: Buyer behaviour is already shifting. A Salesforce 2026 survey reported that 42% of consumers under 35 have used an AI agent for at least one purchase in the past 90 days. In Hong Kong, that demographic skews higher because of high English fluency and high tool adoption.
Force 2: Discovery moves from search to AI conversation. If your products are not indexed by AI agents, you are invisible to a growing slice of intent. Your competitor on Hollywood Road who exposed a clean product feed will be visible. You will not.
Force 3: Brand loyalty weakens at the agent layer. When the AI compares your product to a competitor and the competitor scores 0.3 points higher on review quality, the AI does not feel sentimental about your 28-year history. The data wins.
What HK Retailers Need to Do to Stay Visible
Visibility in agentic commerce is a discipline, not a tool purchase. Five concrete actions cover 80% of what an HK SME owner needs to focus on this year.
The 5-action checklist:
— Publish a clean, structured product feed: schema.org Product markup, accurate stock and pricing, and high-quality images. AI agents read structured data first.
— Make your site agent-readable: clean HTML, no aggressive bot-blocking, robots.txt that allows the major AI crawlers (OAI-SearchBot, ClaudeBot, PerplexityBot).
— Get your products into agent ecosystems: connect to Shopify Merchant Centre, Google Merchant Centre, OpenAI's commerce APIs, and any local Hong Kong marketplaces relevant to your category.
— Treat reviews as ranking signals: AI agents weight third-party reviews heavily. A 4.7 with 200 reviews beats a 5.0 with 8 reviews on most agent ranking models.
— Build a knowledge base for AI agents to query: a public FAQ, a clear returns policy, plain-language product descriptions. The clearer your information, the more often the AI picks you.
Common Misconceptions About Agentic Commerce
Three misconceptions stop Hong Kong SMEs from preparing in time.
Myth 1: "It is only for big retailers." The opposite is true. Agentic commerce levels the playing field for small retailers with great products and clean data, because the AI does not care about brand budget. It cares about product fit, price, reviews, and signal quality.
Myth 2: "It will not affect physical stores." Many agentic transactions trigger physical fulfilment such as same-day delivery, click-and-collect, or in-store pickup. If your inventory data is not online, even your physical store loses agent-driven traffic.
Myth 3: "I can wait until 2028." Agent shopping habits, like search habits in 2010, lock in fast. Customers who learn the new pattern this year will be even less likely to switch back. Late entrants will fight for residual visibility, not new customers.
Agentic Commerce vs Traditional E-commerce: A Practical Comparison
The mental model shift is large. The table below frames the most important differences.
On who decides: traditional e-commerce relies on human discovery and selection. Agentic commerce moves both to the AI, with the human only setting goals and limits.
On what wins the sale: traditional e-commerce rewards strong brand recognition, paid placement, and category positioning. Agentic commerce rewards structured data quality, review signal, and product-fit precision.
On marketing levers: traditional e-commerce uses Facebook ads, Instagram, and Google Search. Agentic commerce shifts spend towards merchant-feed quality, agent-API integration, and generative engine optimisation (GEO).
On time-to-revenue: traditional e-commerce takes weeks to test a new campaign. Agentic commerce changes outcomes within hours of a feed update, because agents query feeds in real time.
Frequently Asked Questions
Q: How much does it cost to prepare my store for agentic commerce?
The base layer (clean schema markup, agent-friendly robots.txt, Merchant Centre connection) typically costs HK$15,000 to HK$50,000 once. Deeper integration with payment-aware agent APIs runs HK$80,000 to HK$200,000. Most HK SMEs can complete the base layer in under a month.
Q: Will agents replace my Shopify or WooCommerce store?
No. They will sit in front of it. Your store remains the fulfilment engine. The agent becomes the new customer-facing layer that drives traffic to it.
Q: Do I need to be on every agent platform?
No. Pick the two or three that match your category and customer base. For HK retail, that usually means Shopify Agent, Google Merchant Centre, and one of ChatGPT or Perplexity.
Q: How is this different from generative engine optimisation (GEO)?
GEO focuses on visibility in AI text answers ("recommend me a sushi place in TST"). Agentic commerce focuses on completed transactions. The two overlap in fundamentals, but the metrics and the integration depth differ.
The Bottom Line for Hong Kong Retail Owners
Agentic commerce is the biggest change to retail discovery since mobile e-commerce in 2012. The good news for Hong Kong SMEs: the playing field is closer to flat than it has been in a decade, because AI agents reward structured data quality over brand budget. The work is unfamiliar, but it is not expensive, and the early movers in your category will compound their advantage every quarter that this technology spreads.
For 28 years, Udomain has helped Hong Kong businesses navigate exactly this kind of platform shift. We understand the cold side of AI, and we understand the hard side of running a small business. Together, we make the technology feel like company, not like another tool.
UD相伴，AI不冷。
Ready to Make Your Store Visible to AI Agents?
If you are running a retail or e-commerce operation in Hong Kong and you are not sure where to start, that is the most common starting point we see. The next move is a 30-minute conversation with our team. We will walk you through every step, from auditing your current product feed to choosing the right agent platforms to structuring your data so AI agents pick you over the competition.
