How shopping chatbots could reshape the future of retail payments

Artificial intelligence is moving rapidly from novelty to infrastructure in retail, and nowhere was that more apparent than at Retail’s Big Show in New York.

The future of retail payments

While retailers once dominated the event, this year the largest crowds gathered around the stands of technology groups promising a future in which shopping is no longer searched for, but delegated.

The rise of so-called agentic AI — autonomous chatbots capable of finding, comparing and even purchasing products — is beginning to challenge long-established assumptions about how online commerce works.

According to estimates from Emarketer, AI-driven platforms will account for around 1.5 per cent of US e-commerce sales this year.

That figure remains small, but the strategic implications are far larger.

Industry executives increasingly view agentic AI as a potential inflection point on a par with the emergence of e-commerce itself in the 1990s or the smartphone revolution of the 2000s.

From search to delegation

For decades, online shopping has relied on consumers doing the hard work: entering keywords, scrolling through pages of results and manually comparing products.

Agentic AI promises to reverse that model.

A shopper might simply describe what they want — for example, a waterproof jacket within a certain budget for a specific trip — and allow an AI agent to source options, weigh reviews and complete the transaction.

Technology companies are racing to become the default interface for this new form of commerce.

Start-ups such as Perplexity and OpenAI, alongside incumbents like Google, want their chatbots to act as digital storefronts, effectively sitting between consumers and retailers.

For users, the appeal lies in reduced friction and cognitive load. For retailers, the consequences are more complex.

A threat to the retail status quo

Agentic AI risks disrupting an online retail ecosystem built over decades at enormous cost.

Control over customer data, search rankings and brand visibility could shift away from retailers towards AI intermediaries.

This is particularly sensitive for companies that rely on “retail media” — the sale of sponsored placements and advertising inventory within digital storefronts — a market estimated by WARC to be worth close to $200bn globally.

If shopping decisions are increasingly made by algorithms rather than humans, the effectiveness of advertising becomes uncertain.

An AI agent may not respond to banner ads or promoted listings in the same way a consumer does, potentially undermining a lucrative revenue stream.

Security concerns also loom large.

Cybersecurity specialists warn that handing personal and payment data to autonomous agents could increase exposure to fraud, manipulated recommendations or malicious code embedded in websites.

Consumer trust, still fragile around AI, remains a significant barrier to adoption.

Walmart advances, Amazon hedges

The diverging strategies of Walmart and Amazon highlight the uncertainty.

Walmart has embraced AI shopping aggressively, hiring senior executives dedicated to the technology and partnering with Google to surface its products within AI-powered search tools.

It has also struck a deal with OpenAI to accelerate its own chatbot capabilities.

Amazon, by contrast, has been more defensive.

While investing heavily in AI infrastructure through its cloud business, it has restricted access to its marketplace for third-party chatbots and is proceeding cautiously with partnerships.

Its own assistant, Rufus, is expanding in functionality but remains tightly integrated within Amazon’s ecosystem.

Analysts note that this approach protects Amazon’s control over customer engagement but may limit exposure to emerging AI-driven traffic.

Payments, trust and the last mile

For payments professionals, the implications are significant.

Agentic AI could streamline checkout, automate payment authorisation and reshape fraud detection. Yet it also raises fundamental questions about consent, transparency and liability.

Who is responsible if an AI makes an unauthorised purchase or is nudged towards a sponsored recommendation without disclosure?

Retailers argue that trust remains their strongest asset.

Loyalty programmes, subscription models and fulfilment capabilities — areas where AI platforms have little presence — may help them retain direct relationships with customers.

Crucially, while AI can recommend and transact, it cannot warehouse, ship or return goods. The physical and financial infrastructure of commerce still matters.

An irreversible shift, but not a total one

Agentic AI is unlikely to eliminate traditional online shopping altogether. As with physical retail, many consumers enjoy browsing and choosing for themselves.

But as AI agents improve, they will increasingly handle routine purchases, subscriptions and replenishment — precisely the areas where frictionless payments matter most.

The retail industry is facing a one-way door moment.

The question is not whether AI agents will play a role in commerce, but how much control retailers and payments providers are willing to cede as the next interface of shopping takes shape.