Image: Supplied
As artificial intelligence (AI) moves from recommendation engines to systems that can act autonomously, the way consumers shop is set to change in fundamental ways. Known as agentic commerce, this next phase of AI promises to delegate everything from product discovery to checkout to intelligent agents operating on a customer’s behalf.
For markets like the UAE, where digital payments, e-commerce adoption and regulatory readiness are already well established, the shift could arrive sooner than many expect.
In this interview with Gulf Business, Daumantas Grigaravicius, head of Middle East at Adyen, explains how agentic commerce differs from today’s AI-driven retail, how close the region is to real-world adoption, and what it means for payments infrastructure, fraud prevention and the future relationship between merchants and consumers.
Agentic commerce is being described as the next evolution of AI in retail. How would you define it, and what makes it different from the AI-driven commerce we’ve seen so far?
Artificial intelligence took a major leap three years ago with the rise of generative AI models, which can process huge amounts of data to answer questions, generate content and support creative and decision-making tasks. The next step in this evolution is agentic AI, which goes beyond simply responding. It can take a complex instruction, reason through it, plan what needs to happen and act autonomously on a user’s behalf.
The intersection of agentic AI and online shopping is where agentic commerce resides, and it is essentially AI doing the shopping for you, from discovery through to checkout and with very limited manual input. Think of it this way: current AI helps you find the perfect pair of trainers by analysing your preferences and past purchases.
An AI agent, on the other hand, would proactively monitor prices across retailers, wait for the optimal moment to buy, apply the best available discounts and complete the purchase – all based on a simple instruction like “buy me the most comfortable pair of running trainers under Dhs500 and have them delivered by Friday”.
What makes this relevant for the UAE is that the foundations are already in place. The Emirates has one of the world’s highest e-commerce adoption rates, consumers are quick to adopt new technologies, and the regulatory landscape has been evolving in a way that makes this shift easier.
Our 2025 Retail Report shows that 70 per cent of UAE consumers use AI tools when shopping – a 44 per cent increase on last year’s figures – which highlights how ready the market is for more autonomous capabilities.
The real game-changer is how agentic commerce will transform the merchant-customer relationship. Instead of competing for attention through traditional channels, merchants will need to ensure their products and services are optimally positioned for agent discovery and selection.
This means creating new touchpoints that work and fit within this autonomous framework while at the same time preserving brand identity and customer loyalty.
How close are we really to AI agents autonomously handling end-to-end shopping experiences in the UAE market? What are the current technical and regulatory barriers?
We’re actually closer than many people think – I’d say we’re looking at adoption that goes beyond pilots within 18 to 24 months for specific use cases, with broader implementation and more sophisticated capabilities following soon after. The momentum is undeniable – new research by cloud software company Salesforce shows that 80 per cent of UAE organisations plan to use AI agents by 2027, up from just 32 per cent today. Right now, we have AI systems that manage personalised product recommendations, dynamic pricing, inventory forecasting and customer support through chatbots.
The next step would be giving these systems more autonomy to act on behalf of customers – for example, searching across multiple e-commerce platforms, accessing and analysing product specifications, reviews and ratings, comparing prices in real time and evaluating return policies and other logistical details before completing the purchase.
From a technical standpoint, the main challenges aren’t prohibitive. We need robust authentication mechanisms to ensure AI agents have explicit authorisation to act on behalf of consumers – what we call verifiable mandates. We’re working with Google and other partners on the Agent Payments Protocol to establish these standards. There’s also the question of interoperability – ensuring AI agents can seamlessly interact with different merchant systems, payment methods and platforms without creating new friction points.
All in all, it’s very encouraging to see genuine collaboration between regulators, technology providers and financial institutions here. The UAE has consistently shown it can move quickly when there’s innovation potential, and agentic commerce certainly qualifies.
As these AI agents begin to make more autonomous decisions on behalf of consumers, how do you see this changing the way people discover and purchase products?
The shift will be significant, but it’s likely to unfold in stages. We’re moving from active shopping to delegated commerce, where consumers define parameters and let AI handle execution. Instead of browsing tens of websites to find the best deal, you’ll set preferences and budgets, then let your AI agent search and place the order.
For UAE consumers, who already value convenience and efficiency, this is a natural evolution. Consider how shopping habits here have already evolved from mall-centric retail to same-day e-commerce delivery.
Agentic commerce takes this further by removing friction entirely from routine purchases. Your AI agent could automatically reorder household essentials when supplies run low, book travel when flight and hotel prices are more reasonable, or even coordinate complex purchases like furnishing a new house within a set budget.
Discovery also becomes far more data-driven in this environment. Instead of being influenced by ads or social media, AI agents will make decisions based on objective criteria – price, quality metrics, delivery times, sustainability scores and more. This puts pressure on merchants to compete on substance rather than just marketing appeal. But here’s what’s crucial – this doesn’t eliminate the human element. High-involvement purchases, luxury goods, experiences – these will still involve personal choice and emotional connection. What changes is that the mundane, repetitive aspects of commerce get automated, freeing consumers to focus on purchases that actually matter to them.
Greater autonomy also raises questions about security and accountability. How do you expect the nature of fraud to evolve, and who would be liable when an AI agent makes an unauthorised purchase?
Fraud in an agentic world becomes both more sophisticated and, paradoxically, more preventable. On one hand, bad actors will try to manipulate AI agents through altered data, prompt injection or by exploiting decision-making patterns. On the other hand, AI agents can apply layers of security that humans simply can’t perform manually – automatic transaction validation, real-time behavioural analysis, instant cross-reference with global fraud databases, and the list goes on.
The key is designing these systems with security built in from the start. Every action taken by an AI agent should have a clear record that the customer approved it – a secure, traceable confirmation that leaves no room for doubt about who authorised what. This creates an immutable audit trail that protects both merchants and consumers. For our part, we’re extending our expertise in tokenisation and authentication directly into agent-led payment flows to ensure these safeguards are robust.
Liability is where clear regulatory frameworks become essential. In our view, there should be a shared responsibility model. The AI platform provider ensures their agent operates within defined parameters and maintains security standards. The merchant remains responsible for delivering goods and services as promised.
The payment processor safeguards the integrity of the transaction and provides the mechanisms for resolving disputes. And consumers maintain responsibility for the mandates and permissions they grant to their agents.
In reality, this shift can help reduce fraud overall. AI agents don’t fall for phishing emails, don’t share passwords and can’t be social engineered. The challenge is ensuring these benefits aren’t offset by new attack vectors, which is why industry collaboration on security standards is so critical.
From a payments and infrastructure standpoint, what advancements do you expect will be needed to fully support agentic commerce at scale over the next few years?
Today’s payment systems are built around human-initiated, discrete transactions. Agentic commerce requires infrastructure that can handle cascading payment flows at massive scale, operating with instant authorisation and settlement to keep pace with AI-driven decision-making.
When an AI agent is comparing prices across hundreds of merchants simultaneously, even the smallest delays could compound into poorer outcomes. The UAE’s advanced digital infrastructure gives us an advantage here, but there’s still work to be done on international transaction speeds.
Token portability becomes crucial. Merchants need to recognise customers seamlessly across agentic channels through universal tokens that work regardless of which AI platform initiates the transaction.
We’re building this into our tokenisation infrastructure – creating identifiers so customers can be recognised easily across different channels. We also need new types of payment instruments designed specifically for AI agents. Think about programmable payment methods with built-in spending rules, multi-signature authorisations for high-value purchases or escrow-like mechanisms for complex transactions. These are evolutionary steps from existing capabilities that need to be standardised.
Perhaps most importantly, we need infrastructure that preserves merchant control and customer choice. The risk of disintermediation is real if we’re not careful. That’s why we’re advocating for and working with partners on open protocols and standards that ensure merchants maintain direct relationships with their customers and own their transaction data, regardless of how AI agents evolve.
The rise of agentic commerce is not a zero-sum game. It builds on technologies already evolving across the ecosystem. And like every major shift in commerce, this new phase is complex, but it’s one we are navigating with our merchants and partners.
For agentic commerce to become sustainable and profitable, we must build infrastructure that delivers genuine trust, transparency and merchant autonomy – because only that way will we achieve outcomes that benefit all.
Read: Retail insights: AI, at your service
