From Checkout to Trust Layer: How Merchants Can Prepare for Agentic Commerce

For more than two decades, online checkout, from the “buy now” button to the credit card prompt to the order confirmation page, has been the heartbeat of digital commerce, not to mention the source of a dopamine rush for shoppers. But as artificial intelligence becomes increasingly autonomous, we’re entering a future where the buyer may not be human at all.

Welcome to the era of agentic commerce, where digital agents act on behalf of people or businesses, making purchase decisions, comparing options and completing transactions automatically. In this world, checkout dissolves into a continuous flow of agent-to-agent interactions.

That shift will redefine how merchants build trust, manage risk and design for shopper intent. It will also have major ramifications for how payments and checkout are handled. The New Challenges of Autonomous Transactions Trust and verification: who’s really buying?

Today’s identity frameworks assume a human triggers the purchase with a password, a biometric, or by clicking the “buy” button in a payment flow. When an autonomous agent executes a purchase, those assumptions no longer hold. Merchants will need to distinguish between the human owner, the software agent and the platform executing the transaction.

That means programmable permissions, verifiable software identities and intent logs that record why and how a purchase occurred — all while keeping the process seamless for the buyer.

Visa recently noted that authentication will evolve beyond Know Your Customer toward Know Your Agent, where merchants must validate both the end user and the autonomous process acting on their behalf. Liability and disputes: If an AI assistant misinterprets an instruction or overspends, who is accountable?

Is it the customer, the merchant or the model provider? Traditional refund and dispute processes weren’t designed for that question. To stay protected, merchants should establish clear contractual terms around autonomous transactions and ensure they have auditable records that explain an agent’s behavior.

In an autonomous environment, transparency replaces the confirmation click: merchants must be able to trace not just what was purchased, but what prompted the agent to make that purchase. The consequences of not fully understanding the purchase intent could be increased chargebacks, higher returns and expensive restocking costs.

Data governance and consent: Agents learn from context, preferences and prior behavior. Every transaction is fueled by data and, therefore, by potential risk. Merchants should treat transparency as part of user experience design.

Customers deserve to know what data their agents can access, what categories they’re authorized to purchase and how that data is used. Privacy is not just a compliance exercise; it’s a differentiator.

According to the Boston Consulting Group, agentic commerce will hinge on merchants’ ability to provide structured, explainable data and clear consent pathways for both humans and their AI agents.

Security and fraud in machine-to-machine commerce: As AI-driven transactions scale, fraud detection must evolve from analyzing human signals to analyzing agent behavior. A compromised agent with stored credentials could execute thousands of micro-transactions before detection.

To mitigate these risks, merchants should tighten API security, encrypt data feeds and implement anomaly detection tuned to agent activity. Payment providers are already experimenting with algorithmic risk scoring, evaluating behavioral patterns rather than passwords. Tomorrow’s fraud may be attributed to stolen code, not stolen cards.

Merchant readiness: Most retailers are still optimizing mobile conversions, not preparing for checkout to fade into the background. Their product data, payment logic and APIs may not be ready for intelligent agents. If catalog data isn’t structured for machine readability, AI assistants can’t interpret it.

If checkout processes aren’t modular and tokenized, autonomous agents can’t transact safely. The readiness gap is as much architectural as it is philosophical. How Merchants Can Prepare Now Make your store machine-readable: Start with the basics: ensure your catalog, pricing, and policy data can be understood by AI systems.

Use structured data standards (schema. org, JSON-LD) and well-documented APIs. Agents will soon crawl product feeds the way search engines crawl web pages, rewarding clarity, consistency and reliable metadata. Within Commerce, solutions like Feedonomics already power this transformation by normalizing product data for AI systems and sales channels alike.

That kind of structured, high-quality data is what will make a merchant “visible” in an agent-driven marketplace. Modernize payments infrastructure. Work with partners that support tokenized, API-first transactions and flexible authentication models. You’ll need the ability to approve conditional or event-based purchases in real time.

Ask your gateway how it plans to manage non-hum