As Agentic Commerce Takes Shape, Payments Face Trust Issues

Artificial intelligence (AI) is already changing how consumers search for products, compare options, and make purchasing decisions. The next step may be less visible but potentially more consequential: allowing AI tools to carry out parts of the purchase journey within instructions set by the customer.

This is emerging against a Malaysian payments market that is already highly digital. In 2025, E-payment transactions grew 25% to 18.4 billion, reflecting how deeply digital transactions have become embedded in everyday commerce.

Often described as agentic commerce, the model could allow AI tools to move beyond helping consumers decide what to buy and support actions such as placing an order or initiating payment within agreed parameters.

For the payments industry, that raises a practical question. If a customer is not manually confirming every step of a transaction, how do merchants know that the purchase still reflects what the customer intended?

According to Eng Sheng Guan, CEO of Fiuu, the fundamentals of payments do not change simply because more of the purchase journey becomes automated.

“Customers still need confidence that every transaction is authorised, secure and reliable. As AI plays a greater role in facilitating purchases, the important thing is ensuring that it acts with the customer’s permission and within clearly defined limits.”

Permission becomes part of the payment journey

In a conventional online transaction, consumers typically move directly through each stage, from choosing a product to confirming payment.

Agentic commerce could change that sequence. A customer might, for example, instruct an AI tool to purchase a particular item within a set price range, or book a service if certain timing, and cost conditions are met.

This makes the scope of the customer’s permission increasingly important.

“The first consideration should be customer permission,” Eng said. “Customers need to understand what they are allowing an agent to do and have the ability to change or withdraw those permissions for future purchases.”

He added that identifying the customer and obtaining payment approval are important, but neither necessarily establishes the full scope of what an AI tool has been authorised to do.

That distinction becomes important when circumstances change. If a product exceeds the agreed price, a preferred option is unavailable, or the terms of a purchase materially change, businesses may need to determine whether the transaction can still proceed or whether the customer should be brought back into the decision.

Keeping the experience simple

Payments have spent years moving towards faster and more seamless checkout experiences. Greater automation should not reverse that progress by adding unnecessary friction.

Instead, the challenge will be making the safeguards around automated purchases increasingly intelligent and contextual.

This could involve clearer signals around authorised activity, secure payment credentials, consistent risk controls, and reliable transaction records. Much of that complexity should ideally remain in the background, so the customer experience stays simple.

The same principle applies when a transaction requires attention.

Eng added, “As more of the purchase journey becomes automated, payments should still feel straightforward to the customer. If something goes wrong, there also needs to be a clear route to assistance with merchants, platforms, and payment providers understanding their respective responsibilities, and having a reliable record of what happened.”

This is particularly relevant where multiple parties may be involved in a single purchase journey. Automation may make the transaction easier, but it should not make it harder for customers to understand what they authorised or resolve an issue afterwards.

Where could businesses see this first?

Agentic commerce is still evolving, so businesses do not need to overhaul their payment environments yet. A more practical starting point is to consider where greater automation could realistically fit within the customer journey.

Repeat purchases may be one area where these models begin to emerge. Travel is another. A customer could, for instance, specify an acceptable price, preferred timing, and alternative options for a service such as an airport transfer.

The important question is where the boundary sits.

Some changes may remain comfortably within the customer’s original instructions. Others may require a fresh decision. Establishing that boundary will help businesses determine how much of a purchase can responsibly be automated while keeping the customer in control.

A regional interoperability question

For Southeast Asia, agentic commerce also raises a wider issue.

Consumers across the region use different combinations of cards, digital wallets, bank transfers, QR systems, and domestic payment schemes. Approval requirements and regulatory expectations also differ between markets.

As businesses increasingly serve customers across borders, more automated commerce will need to operate within that diversity.

The question is therefore not only whether AI can support more of the purchase journey, but whether the underlying payment ecosystem can do so consistently across different methods and markets.

“The long-term value of agentic commerce will not come from automation for its own sake. It will depend on whether we can make these new experiences easier to support while keeping customer permission, control, and payment security clear throughout the transaction,” Eng elaborated.

As agentic commerce takes shape, that balance may prove more important than speed alone.

#businessnews