Is Your MENA Payment Stack Ready for AI Checkout? Agentic Commerce in MENA is here

Imagine a shopper in Riyadh. She asks her AI assistant to reorder her monthly skincare routine: same brands, within budget, fastest delivery. The agent searches, selects a merchant, and attempts checkout. Your payment gateway returns an error it cannot classify. The transaction is declined. The shopper never sees an error message. She just never gets the order. As a merchant, you never knew the sale was there to lose.

That scenario is not fictitious anymore. It is happening right now, at low volume, across every market where AI shopping assistants are in use, and volume is building fast. ChatGPT's Instant Checkout has been live since September 2025. Google announced its Universal Commerce Protocol (UCP) in January 2026, with Walmart, Target, Shopify, and more than 20 other major retailers already committed.

McKinsey's QuantumBlack research (October 2025) projects that agentic commerce will orchestrate between $3 trillion and $5 trillion in global retail spend by 2030. Juniper Research's April 2026 forecast places agentic spend at $8 billion in 2026 alone, rising to $1.5 trillion globally by 2030.

The question for MENA merchants is whether your payment infrastructure is built to handle it when it arrives at scale. In the GCC, scale tends to arrive faster than anyone planned for.

Key Takeaways on agentic commerce in MENA

• Agentic commerce describes transactions completed by AI agents on a user's behalf, without a human clicking through checkout. ChatGPT's Instant Checkout has been live since September 2025.

• Agent-initiated transactions carry different authentication signals than standard e-commerce transactions. Most MENA payment stacks are not currently configured to handle them correctly.

• Saudi Arabia reached 79% non-cash retail transactions by Q1 2025, ahead of its Vision 2030 target. GCC consumers adopt new payment behaviours faster than most markets.

• A provider-agnostic secure vault, no-code routing rules, and multi-PSP orchestration are the three infrastructure requirements that determine whether a merchant is ready for agentic commerce.

• Merchants with higher checkout completion rates are more likely to be selected and recommended by AI shopping agents. Payment infrastructure performance is becoming a distribution channel decision.

• Apaya is a no-code payment orchestration platform built for MENA merchants. Its routing layer, secure vault, and multi-PSP connectivity directly address the infrastructure requirements described in this article.

What agentic commerce actually is

There is a lot of noise around AI in payments right now, so it is worth being specific.

Agentic commerce is a form of e-commerce in which an AI agent, acting on a user's behalf and within limits the user has set, searches, evaluates, and completes a purchase without the person clicking through a traditional checkout flow. The buyer is still accountable. The initiator is no longer human.

That distinction matters enormously for payment infrastructure, because everything your checkout currently assumes needs revisiting: that a person is present, that authentication happens in real time with a human responding, that the transaction is initiated by a specific device at a specific moment.

The practical effect on your payment stack is immediate. An agent-initiated transaction looks different from a standard e-commerce transaction. It may carry different authentication signals. It may use a stored credential rather than a card entered at the moment of checkout. It may fail fraud rules that were built to filter bot traffic, because a well-designed AI agent and a bot can look remarkably similar at the infrastructure level.

Why MENA merchants face this earlier than they think

The GCC has a track record of compressing adoption timelines. Digital wallets went from niche to mainstream faster here than in most Western markets. Saudi Arabia surpassed its Vision 2030 cashless target of 70% non-cash retail transactions by Q1 2025, ahead of the government's own deadline, according to the Saudi Central Bank.

Dubai recorded 88% cashless usage under its Cashless Strategy. The UAE Instant Payment Platform captured 28% of domestic transfers within six months of launch, according to the MENA Digital Payments Market Report 2026 (Mordor Intelligence, March 2026).

When new payment behaviours arrive in the GCC, they move quickly. That is partly infrastructure, partly demographics, and partly government intent. The region's population skews young, smartphone penetration is high, and regulators are actively incentivising digital adoption rather than tolerating it reluctantly.

AI assistant usage follows the same pattern. GCC consumers are not late adopters of technology. The merchants serving them need to be ready before volume arrives. Not after.

"Every merchant we speak to in the GCC understands that AI is changing how their customers shop. If an AI agent cannot complete a purchase reliably, it stops trying. Your checkout performance is about to become a distribution channel decision." Sally Hanekom (Head of Marketing, Apaya)

There is also a MENA-specific fragmentation problem. Your payment stack likely spans multiple PSPs: a global provider like Checkout or Stripe for international cards, a regional acquirer for Mada or KNET, possibly a BNPL provider like Tabby or Tamara. An agent-initiated transaction needs to navigate that stack correctly. Without an orchestration layer, there is no mechanism to identify the transaction type, apply appropriate routing logic, or set different fraud thresholds for agent traffic versus standard checkout traffic.

A single-PSP setup is even more exposed. If the PSP cannot interpret an agent-initiated request correctly, or if it applies a blanket fraud block on non-standard authentication signals, the transaction fails silently. You lose the sale and the data that would tell you why.

The infrastructure gap most merchants are not thinking about

Payment infrastructure has been built on one assumption that is now being tested: a human is present at the moment of authorisation.

PCI DSS, card network rules, and authentication standards, including 3D Secure, all define roles and responsibilities for a payment flow where someone is actively completing a checkout. They do not define what happens when software is acting on that person's behalf, carrying credentials that the user pre-authorised.

This creates a real operational gap. An agentic transaction may present as card-not-present, use a stored token, carry no 3DS response, and originate from an IP address belonging to an AI infrastructure provider. To a fraud system built on traditional signals, that pattern looks suspicious. The transaction gets declined. The agent cannot retry through a different provider because there is no orchestration layer to make that call.

The merchants most exposed here are those running a single PSP without a routing layer, relying on their PSP's built-in fraud tools, and without a provider-agnostic token vault. That description covers most mid-market MENA merchants today.

The best-positioned merchants are those who have already separated their payment infrastructure from any single provider's logic. They can apply routing rules based on transaction type, set different fraud parameters for different traffic sources, and update those rules without writing code.

What your payment stack needs before this becomes mainstream

You do not need to rebuild your checkout for agentic commerce. You need your existing infrastructure to be flexible enough to handle a new type of transaction without failing or blocking it. Here is what that looks like in practice.

1) Identify agent-initiated transactions at the point of entry. Emerging protocols, including OpenAI's Agentic Commerce Protocol (ACP) and Google's Universal Commerce Protocol (UCP), are building standardised signals into agent-initiated requests. Your payment stack needs to be able to read those signals and route accordingly. At present, this means ensuring your orchestration layer is updated as these protocols mature, rather than waiting on your PSP to interpret them.

2) Apply separate fraud rules for agent traffic. The fraud signals for an AI agent completing a legitimate purchase look different from the signals for a human doing the same thing. If your fraud rules treat them identically, you will block genuine agent transactions. If you have no way to distinguish between them, you have no way to tune your rules. Orchestration gives you that separation: you can set agent-initiated transactions to a different threshold, monitor them separately, and adjust in real time without touching your codebase.

3) Ensure your token vault is provider-agnostic. Agentic transactions depend on stored credentials. If those credentials are locked inside a single PSP's vault, an agent cannot reference them when routing through a different provider. That means a failed transaction cannot be retried through a better-suited acquirer. The answer is a secure vault that sits outside any single PSP's infrastructure, where encrypted credentials are stored in a way that keeps them portable. When your vault is genuinely provider-agnostic, your orchestration layer can route freely without hitting a credential portability ceiling. Without that separation, token lock-in becomes the hard limit on everything else you are trying to build.

4) Route without engineering intervention. When something changes, whether a new protocol launches, a PSP performs poorly on agent traffic, or a fraud pattern emerges, you need to be able to update your routing rules in minutes, not weeks. No-code orchestration is not just a convenience here. It is the operational speed you need to adapt to a market that is moving faster than traditional development cycles allow.

5) Maintain visibility across all transaction types in one place. If your agent-initiated transactions are running through a different system from your standard checkout traffic, you will not be able to see patterns across both. Unified analytics, showing approval rates, decline reasons, and fraud flags across all PSPs and all transaction types, is the only way to know whether your agentic payment setup is performing or silently failing.

The competitive angle that most MENA merchants are missing

There is a second dimension to this that goes beyond infrastructure readiness.

AI shopping agents do not select merchants randomly. They select based on structured product data, pricing clarity, delivery reliability, and, critically, checkout completion rate. An agent evaluating two merchants selling the same product will bias towards the one with a higher historical checkout success rate, because a failed checkout is a failed task.

Your payment infrastructure's performance is becoming a factor in whether AI agents recommend your business at all.

"We have been watching the adoption curve on agentic commerce closely, and what strikes me is how quickly the GCC moves once a behaviour takes hold. We saw it with contactless, with BNPL, with digital wallets. Merchants who prepared their infrastructure before the inflection point captured significantly more of that volume than those who scrambled to catch up." Sally Hanekom, Head of Marketing, Apaya

That is a genuinely new dynamic. For the past decade, checkout optimisation has been about the human experience: reducing friction, building trust, offering local payment methods. Those things still matter. But a new layer is being added on top, namely machine-readability. Can an AI agent successfully complete a purchase with you? If not, your conversion rate suffers not just with that agent, but potentially with every user of that agent going forward.

Merchants who get their payment infrastructure right early, who can handle agent-initiated transactions, who have high approval rates across all transaction types, and who present structured, reliable checkout signals, are building a genuine competitive advantage in a channel that will be significant by 2027.

What to do in the next 90 days

You do not need a six-month infrastructure project to prepare for agentic commerce. You need a clear picture of where your current setup would fail, and a plan to close those gaps before they cost you measurable revenue.

Start with these questions:

1) Can your current payment setup identify and separately route a non-human-initiated transaction? If the answer is no, you need an orchestration layer that can.

2) Are your customer payment credentials stored in a vault you control, or inside a single PSP's infrastructure? If it is the latter, your tokens are not portable. Agent retries through alternative providers will fail.

3) Can your fraud team update transaction-level rules without raising a development ticket? If not, you will not be able to respond quickly enough when agent traffic patterns emerge.

4) How long would it take you to add a new PSP or payment method to your checkout if a specific provider started performing better on agent traffic? If the answer is weeks or months, you are already behind.

The merchants who will capture the most value from agentic commerce are not necessarily the ones with the most sophisticated AI strategy. They are the ones whose payment infrastructure is flexible enough to handle whatever comes next, without stopping to rebuild it first.

AI shopping agents are not replacing your customers. They are acting on behalf of them. And right now, most MENA merchant payment stacks are not built to handle that distinction. The good news is that the infrastructure you need to be ready is the same infrastructure that makes your payments better across every other dimension: multi-PSP orchestration, provider-agnostic token vaulting, no-code routing rules, and unified analytics across all transaction types. Getting ready for agentic commerce is not a separate project. It is an acceleration of what forward-thinking MENA merchants are already doing.

Frequently Asked Questions

What is agentic commerce? Agentic commerce is a form of e-commerce in which an AI agent acts on a user's behalf to search, evaluate, and complete a purchase autonomously. The user sets parameters, budget, brand preferences, delivery requirements, and the agent executes the transaction without the person manually clicking through a checkout flow. ChatGPT's Instant Checkout and Google's Universal Commerce Protocol (UCP) are the two most significant live implementations as of 2026.

How does agentic commerce affect payment infrastructure? Agent-initiated transactions carry different authentication signals from standard human-initiated transactions. They may present as card-not-present, reference a stored token, and carry no 3DS response. Payment infrastructure that assumes a human is present at the moment of authorisation will either decline or misroute these transactions. Merchants need multi-PSP orchestration, provider-agnostic token storage, and the ability to apply separate fraud rules for agent traffic to handle them correctly.

Are MENA merchants at risk from agentic commerce earlier than other markets? Yes. The GCC has a consistent track record of compressing payment adoption timelines. Saudi Arabia surpassed its own cashless targets ahead of schedule. The UAE Instant Payment Platform captured 28% of domestic transfers within six months of launch. When new payment behaviours take hold in the GCC, they scale quickly. Merchants serving GCC consumers need to prepare infrastructure before volume arrives, not after.

What infrastructure does a merchant need to support AI-driven checkout? Four capabilities are required: the ability to identify and route agent-initiated transactions separately from standard traffic; separate fraud rules for agent traffic; a provider-agnostic secure vault where encrypted credentials are stored portably; and no-code routing rules that can be updated in real time without engineering involvement.

Does payment infrastructure performance affect whether AI agents recommend a merchant? Yes. AI shopping agents evaluate merchants partly on checkout completion rate. An agent that encounters repeated checkout failures with a given merchant will deprioritise that merchant in future recommendations. Payment performance is therefore becoming a factor in AI-driven discovery and distribution, not just in transaction completion.

What is payment orchestration, and how does it relate to agentic commerce? Payment orchestration is a software layer that connects a merchant to multiple payment service providers (PSPs) through a single integration. It handles routing logic, fallback behaviour, fraud rule management, and transaction analytics across all connected providers. For agentic commerce, orchestration is the enabling layer that allows agent-initiated transactions to be routed correctly, retried intelligently, and monitored separately from standard checkout traffic.

How does Apaya help merchants prepare for agentic commerce? Apaya is a no-code payment orchestration platform built for MENA merchants. Its routing layer supports conditional logic by transaction type, its secure vault stores encrypted credentials independently of any single PSP, and its no-code Studio allows payments and operations teams to update routing rules in real time without raising development tickets. Merchants using Apaya can configure their stack to handle agent-initiated transactions without a separate infrastructure build.

About Apaya

Apaya is a no-code payment orchestration platform purpose-built for MENA merchants. It connects businesses to 60+ payment service providers through a single integration, with intelligent routing, a provider-agnostic secure vault, and a drag-and-drop Studio that allows non-technical teams to manage payment workflows without engineering dependency. Apaya serves merchants across the UAE, Saudi Arabia, Egypt, and the wider GCC.

Want to understand how your current payment stack handles agent-initiated transactions? Book a 30-minute call with our team , and we will walk through it with you.