Agentic AI is about to rewrite how we shop

Trust, not tech, will decide the future of AI shopping agents By Jenny Hadlow, Chief Operating Officer, Checkout.com

For years, retail has focused on shaving a click or two off the path to purchase. Agentic AI – systems that act on a consumer’s behalf rather than merely recommending options – is about to change the game entirely. Instead of trawling websites for the best deal, shoppers will increasingly brief an AI: “Find my partner a thoughtful gift under £75, delivered before Friday,” or “Keep my household essentials stocked at the best price.” The agent does the searching, comparing and, before long, even the transacting.

Checkout.com’s research across the UK and US shows this shift is already underway, especially in the run-up to Christmas when gift-buying stress peaks. But it also highlights a crucial point for retailers: the biggest barrier is not AI itself. It is the fear of losing control, and of heightened security and fraud risks. Trust and control must become core product features if retailers want to win in the era of agentic commerce.

Consumers are already using AI to shop – especially for gifts

AI has already become part of everyday shopping for a significant portion of consumers. Almost half say they have used it to help choose a gift for a partner, with usage rising dramatically among those in their mid-twenties to mid-thirties. Around a quarter now use AI to compare product reviews, and nearly as many use it for general gift inspiration. Many also rely on AI for practical help, whether that means finding discount codes, checking stock availability, or identifying the fastest delivery options. Increasingly, consumers see AI as a budgeting tool: a third use it to secure the best price, and another third use it to help control spending.

Yet even as AI becomes embedded in everyday browsing and buying, consumers hesitate to let it take the next step and transact autonomously. When looking ahead, shoppers believe that within five years, AI could handle around 21% of their monthly spending. But today, more than half say they do not feel comfortable with an AI agent making a purchase for them. The future they can imagine and the present they trust are still some distance apart.

The blockers: control and security, not the idea of AI

Most consumers are not opposed to AI in principle. A majority say it is entirely acceptable to use AI for gift ideas or general shopping guidance, and those who already do so feel overwhelmingly comfortable admitting it. The hesitation comes from the sense of losing control. Many worry they will not have oversight of what is purchased, or that an AI will not understand their preferences well enough to make the right choice. Others fear being accidentally signed up to subscriptions or overspending without realising.

Layered onto this is a deeper unease around security. Concerns about fraud, hacking and data misuse are widespread. They sit within a broader context in which consumers already distrust online marketplaces, citing counterfeit goods, misleading reviews, insecure checkouts and opaque data practices as long-standing issues. It is hardly surprising, then, that a technology designed to spend money on their behalf triggers caution.

Where shoppers actually want AI to step in

While many consumers hesitate about fully autonomous AI shopping, they are far more open to delegating repetitive, low-stakes tasks. Nearly half say they would use an AI agent to make “boring” purchases for them. When asked which tasks they would realistically hand over, shoppers point to everyday household items (18%), weekly food shops (15%), Secret Santa obligations (12%), travel bookings (12%) and pet supplies (12%). These categories lack the emotional weight of gifting or fashion and therefore offer ideal testing grounds for agentic commerce.

If AI can reliably save time, reduce cognitive load and keep spending on track in these foundational areas, it will earn the right to take on more complex categories later. Consumers are also clear about what they need before granting that trust. They want the ability to set spending limits (28%); real-time alerts for every purchase (24%); higher-value items to require human approval (23%); easy, transparent refunds and returns (30%); and above all, they want to know the system is secure (31%). These expectations form not barriers but the blueprint for how agentic commerce must evolve.

What retailers need to build for agentic commerce

To meet these expectations, retailers will need to rethink both the customer experience and the underlying systems that will increasingly serve not just human shoppers but the AI agents acting on their behalf. If agents are the new shoppers, merchants must create environments that both the machine and the human behind it can interpret and trust.

Machine-readable catalogues and real-time pricing APIs will become critical, enabling agents to understand availability, compare options instantly and avoid uncertainty that leads to abandoned baskets. Retailers will also need to make trust visible. Clear delivery windows, transparent refund and returns policies, and simple explanations of how decisions are made will help both humans and AI agents navigate purchases with confidence.

Just as important is the payments layer. Secure, resilient payment flows and instant refunds will become competitive differentiators because agents will optimise for confidence, not clicks. They will choose merchants who minimise risk, make decisions reversible and spell out policies explicitly. All of this must be underpinned by strong consumer controls. Shoppers want limits, alerts, approval steps and reliable returns — features many explicitly say would increase their comfort with AI-driven purchases. Above all, they want assurance that the system is secure.

From recommendation engines to responsibility engines

Retail’s digital transformation over the past decade was built on personalisation and convenience. Agentic commerce raises the stakes dramatically. These systems are no longer just suggesting products; they are choosing and purchasing them. That shift demands a new level of responsibility.

The winners will be those who understand that agency must be earned. They will prioritise transparency, security and shared control. They will use AI to support and protect customers, not push them toward transactions they cannot oversee. And they will build momentum gradually, proving reliability in low-stakes, everyday tasks before moving into more emotionally charged categories like gifting.

Consumers are ready for help, but convenience alone is not enough. They want confidence, clarity and control. Retailers who deliver all three will turn AI agents from a novelty into a trusted co-pilot – one that makes life easier while keeping the customer firmly in the driving seat.