Preparing eCommerce for the Agentic Commerce Era

(based on US Census data and forecasted adoption rates). It is vital to understand how to adapt to the new future of AI shoppers to avoid losing out on potential sales and revenue.

Agentic commerce can be described as Business-to-Agent (B2A) shopping. It allows merchants to sell products 24/7 to customers who are too busy to browse their ecommerce platform. A customer can tell the AI agent, "help me determine the best subscription tier for this app based on my usage", and it will do that within seconds, providing the best pricing option for this specific user.

However, to ensure AI agents can make frictionless purchases, a digital storefront's infrastructure will need to be able to talk directly to that AI. With the right partners, infrastructure and standards in place, a new sales channel will open up where the shopper is an algorithm that doesn't get distracted or abandon its basket.

Build an AI-ready payment experience

Preparing for the New Payment Ecosystem

In the current payment ecosystem, an ecommerce platform is designed for humans - focused on a smooth user interface and brand loyalty. But an AI agent does not care about branding or how nice a website looks. It cares about whether the transaction goes through. It won’t be swayed by loyalty or an efficient UX. Agentic commerce is algorithmic, not emotional, which means in this new payment ecosystem, the strategy for merchants is to adopt machine-to-machine commerce. Sell to systems, not to humans.

For digital-native, multi-market businesses, a strong user experience (UX) has always been a top priority for converting potential customers into buyers. It led to a reduction in the number of browser windows that customers need to navigate when making an online purchase (as each new window is another opportunity for them to abandon the trolley). With AI agents handling the shopping, the existing retail infrastructure is no longer optimal; however, this shift allows merchants to drastically reduce the cost of conversion. By optimising for AI agents the path to revenue is straight, automated, and hyper-efficient.

In the new payment ecosystem, driven by agentic commerce, the strategic shift is about optimising the payment experience (PX). The parts of commerce that used to be UX problems are becoming protocol problems, as a poor checkout interface won’t deter an AI agent, whereas a faulty transaction and poor conversion process will.

For agentic commerce to work, ecommerce platforms need to optimise the purchase and payment flows for AI agents. Agentic commerce introduces a new player acting on behalf of the consumer, which brings potential risks and complexity — who is liable and accountable when something goes wrong? It is therefore critical to build assurance into a transaction that an AI agent is acting genuinely on behalf of the consumer, and to make agentic payments auditable — creating a clear, verifiable trail of authorisation and intent. Without this foundation of trust, agentic commerce will not scale.

To achieve this, digital merchants and their partners need to understand each component of this new digital ecosystem:

• Agent-ready commerce stack: The first step is ensuring product data is structured in a machine-readable format. An AI agent is not reading the text on the store; it is reading the metadata and the APIs. Merchants need to allow agents to manage subscriptions and billing autonomously based on real-time signals. The backend of an ecommerce platform should be viewed as the storefront made for AI, and if an agent can’t see the price or terms of a product in a structured data format, then it can’t make a purchase.
  

Further ahead, some merchants may want to launch their own agents to interact with other agents — enabling the AI to negotiate prices and manage usage-based billing.

Create trust through verified agent authority

• Agent verification: Merchants should ensure that only verified agents can access their catalogue and initiate purchases. This will prevent unauthorised and malicious bots from committing fraud, and bolster security for the ecommerce platform. In practice, this may mean requiring an authentication proof of intent when a transaction request arrives, ensuring there is a verifiable link between the purchase and the consumer’s original instruction.

• Delegated authentication: Agentic commerce can take different forms depending on whether the consumer is present at the moment of purchase or has delegated purchasing authority to an AI agent in advance, within certain defined limits. Delegated authentication is key, as generative AI agents are, by design, non-deterministic and therefore need guardrails set by their users. These guardrails are equally important for merchants, as they provide proof that the customer initiated or authorised the transaction, for example in a chargeback or dispute situation.
  

Examples of where delegated authentication is required include:

• Monitoring prices or availability and purchasing when predefined conditions are met.
• Repeat purchases based on past behaviour.
• Complex automation request, e.g. "If my flight gets delayed over 3 hours, rebook me on the next available route and update my hotel check-in".
• Don’t neglect your human customers: While AI agents don’t require an emphasis on traditional UX and branding, that does not mean that those areas can be neglected. Having an AI agent ready ecommerce platform still requires a high-quality, easy to navigate catalogue of items that provides clarity and competitiveness. If a human can’t find details like pricing and product availability, then an AI certainly will not.
  

Ultimately, it comes down to ensuring the customer knows the risks of using an AI agent for their shopping, and ensuring the internal infrastructure of the ecommerce platform and payment process is easily understood by AI agents. Within this new payment ecosystem, merchants must protect themselves and not make mistakes that impact consumer trust.

Turn agentic commerce into practical growth

Practical uses for digital businesses

Agentic commerce is particularly relevant to digital goods because they are standardised. While a person might need to look at 20 different shirts before deciding on their preferred choice, a shopper can easily tell a bot to find the best price for a specific music streaming tier or a new game release. It does not require the customer to remember to buy a specific product and then go out of their way to find it; the AI agent can respond to its user’s needs as soon as the moment arises.

As agentic commerce is still early, its use cases are still emerging - future-facing ideas rather than proven use cases. Here are some examples of what agentic commerce may be able to do in the future for digital retailers:

• In a video game with micro-transactions, an AI agent can be set up to make quick purchases based on the player’s in-game status, within a predetermined allowance. If a player is running low on stamina or runs out entirely, the AI agent can automatically purchase a refill without the player ever having to pause the game or open a menu. It reduces the likelihood that the player will drop off due to a clunky in-game menu system or game downtime.

• For digital content platforms, agentic commerce allows customers to find a very specific product, such as the director’s cut of a film or a digital art book of a show, for a very specific price. It can handle this in two ways: either by searching several stores across the entire web until it finds one that meets those specific criteria, or by attempting to negotiate with your store's own AI agent. An AI sales associate could offer a discount to a shopper to entice the AI agent to make a purchase at this store, rather than go to a different storefront, all set within your own predetermined rules of what is allowed.

• For stores and software that offer subscription services, an AI agent can dynamically subscribe and unsubscribe based on specific scenarios, like needing a music subscription for a party. For the merchant, this dynamic upselling can keep customers from sticking to low-tier plans and capture spend from specific high-value moments that would otherwise be lost.

Control risk without slowing innovation

Managing the new risks

Autonomous agents change how transactions are initiated, requiring payment infrastructure that can manage new operational risks. These challenges center on agent security, intent verification, spending controls, and automated resolution frameworks.

• Managing security and malicious agents: To mitigate risks from malicious actors, merchants must implement identity verification protocols like Know Your Agent (KYA). By transacting with verified agents, businesses ensure digital actors are authenticated by recognized issuers. This verification allows for the implementation of blacklisting and whitelisting services, helping to shift financial liability away from the merchant if unauthorized agent behavior occurs.

• Verifying consumer intent: In the European market, matching a customer's original intent to the final transaction is a legal and technical necessity. Infrastructure must securely record and verify this intent—such as specific price caps or product requirements—throughout the transaction lifecycle. Purpose-built purchase intent verification ensures the final settlement accurately reflects the human instruction, reducing the risk of agents misinterpreting complex requests.

• Controlling scope and credentials: Merchants need systems to validate that agents remain within their delegated mandates. This involves real-time checks against digital contracts that define a consumer’s spending power, budget limits, and vendor preferences. Additionally, evolving strong customer authentication (SCA) for an agentic environment is necessary to protect payment credentials while allowing for seamless automation.

• Resolving disputes and chargebacks: Transparent evidence layers are required to reconcile machine-to-machine transactions at scale. By linking cryptographically verified intent to the final settlement, merchants can prove an agent acted within its pre-defined mandate, which narrows the valid grounds for chargebacks. Standardized data protocols enable these conflicts to be settled through neutral, automated processes rather than manual intervention.

• Technical readiness and stability: High-volume agent activity can strain checkout systems, particularly during peak periods like flash sales. Merchants should implement robust rate limiting to ensure their digital infrastructure remains stable. This prevents a sudden influx of automated traffic from crashing the store and ensures agents do not accidentally initiate duplicate transactions.

Prepare for evolving agentic payment standards

Emerging Industry Protocols

Each payment provider has a different system infrastructure and so the provider and merchant must work together to properly implement the right payment protocols, to enable safe agentic commerce.

There are various emerging industry standards and protocols aiming to enable trusted agentic commerce. Worldline is collaborating with industry leaders to enable the key agentic commerce protocols for merchants..

• Model Context Protocol (MCP)
  The MCP serves as a secure bridge and translation layer between Large Language Models (LLMs) and agentic ready APIs. A protocol like this is required to enable the AI agents to initiate payment actions via customer prompts. These agent-initiated actions include:
  • payment creation
  • refunds
  • status checks
  • payment capture
  • share secure payment links while maintaining security and compliance