Your next customer might be an AI agent. Is your storefront ready?

a person shopping online — Consumers are starting to hand over their shopping to AI. (Source: Polina Tankilevitch via Pexels)

By Vercel April 15, 2026

Text Size

A quiet shift is underway in retail, and most online stores are not built for it.

Consumers are starting to hand over their shopping to AI. Not in some distant, theoretical future, but now. Half of all Australians have already used AI assistants for online shopping, according to PayPal research released in 2025. Among shoppers under 45, that figure rises to two-thirds. More telling still, 78 per cent of Australians expect AI assistants to become a mainstream part of the shopping journey, with over half planning to use them.

This is agentic commerce: AI agents acting on behalf of shoppers to discover, evaluate, and purchase products with minimal human input. Asking ChatGPT to compare running shoes and letting AI assistants like Alexa reorder household staples. Delegating gift research to tools that can scan dozens of retailers in seconds and come back with a shortlist.

And it’s a global phenomenon. Adobe reported a 4700 per cent year-over-year increase in traffic to US retail sites from agentic shoppers last year. Meanwhile, BCG research shows customers arriving via AI agents are 10 per cent more engaged than traditional visitors, reaching retailers further down the sales funnel.

The message for Australian retailers is clear: AI agents are already moving shoppers through the funnel. The only question is whether your store is visible when they do.

What agents need from your storefront

An AI agent doesn’t browse a website the way a person does. It doesn’t scroll through hero banners or respond to promotional pop-ups. It queries APIs and reads structured data. It needs fast, reliable responses to questions about inventory, pricing, product specs and shipping, often across multiple retailers at once.

That means the storefront of the future needs to be machine-readable, not just human-friendly. Retailers whose digital infrastructure is locked into monolithic platforms, or whose product data is buried in page code, will simply be invisible to this new class of consumer.

The retailers who win will be those with fast, composable architectures that expose clean data to AI agents while still delivering great experiences for human shoppers.

Vercel has been partnering with retailers such as David Jones, Petbarn, Bombas, and Carhartt for years to deliver lightning-fast composable commerce. We’re now unlocking the power of agentic infrastructure for their customers at the bleeding edge. As you develop an agentic commerce strategy, here are the first five things we recommend you consider:

Speed that agents demand: Agentic interactions happen in real time. An AI agent querying inventory and pricing across a dozen retailers has zero tolerance for slow responses. Vercel’s global CDN and fluid compute infrastructure deliver the performance and reliability that both human visitors and agents require at scale.

Composable by design: Agentic commerce favours retailers with composable stacks – storefronts assembled from best-of-breed headless components: CMS, commerce engine, search, and payments. This architecture separates the presentation layer from the data and transaction layer, making it far easier to expose clean, structured APIs to AI agents.

Tools to build AI-native experiences: Retailers need a toolkit to build agentic commerce experiences directly: In-site shopping assistants, conversational interfaces, and agentic checkout flows. Vercel’s AI SDK and AI Gateway give retail teams the industry-leading toolkits they need, and the Vercel Marketplace also offers a dedicated AI Agents & Services category with pre-built agents, unified billing, and native observability.

Commerce beyond the storefront: As AI shopping shifts into chat and messaging apps, retailers need their agents to meet customers where they already are. Vercel’s Chat SDK lets teams build agent logic once and deploy it across Slack, WhatsApp, Microsoft Teams, and other platforms. That means a shopping assistant built for your website can also run in a customer’s preferred messaging app, with no need to rebuild for each channel.

MCP support for structured agent access: Model Context Protocol (MCP) is the emerging standard that lets AI agents fetch live retailer data directly, rather than scraping rendered pages. Vercel enables teams to build and deploy MCP (Model Context Protocol) servers, making structured, agent-to-retailer interaction possible at scale.

The competitive window is closing

The demand side is moving fast. The supply side needs to catch up.

Globally, 63 per cent of retailers agree that companies without AI agent capabilities will fall behind within two years, according to Deloitte. Australian retailers face that same pressure, but with a consumer base that’s less patient than most.

The retailers who move now, building storefronts that are fast, composable, and machine-readable, will be the ones AI agents recommend, transact with, and return to. Everyone else risks becoming invisible to an entire new channel of commerce.

The shift to agentic commerce isn’t coming. It’s here.

The question is whether your storefront is ready for it.