AI Agents Are Your Next Customer and Your Checkout Isn’t Built for Them

Retailers have spent the last decade obsessing over conversion rates, checkout optimization, and reducing friction for human buyers. However, a new customer has quietly entered the market, one that doesn’t browse the way humans do, doesn’t type into forms, and doesn’t tolerate broken workflows. Who are they? Artificial intelligence agents. According to Sean Neville, cofounder of Circle and architect of USDC; CEO of Catena Labs, "The bottleneck for the agent economy is shifting from intelligence to identity. In financial services, 'non-human identities' now outnumber human employees 96-to-1, yet these identities remain unbanked ghosts."

AI Agents Are Already Doing the Work

AI agents are already researching products, comparing prices, managing replenishment, and making recommendations on behalf of consumers and businesses. Agentic commerce is no longer theoretical. It’s happening now. Yet when agents reach the point of purchase, transactions often fail. This isn't because the agent lacks intent. It's because the retail infrastructure still assumes a human is the buyer behind the keyboard.

Checkout is Broken

Today’s checkout flows are fundamentally built for humans. They require logins, form fields, CAPTCHA challenges, manual entry, and step-by-step confirmations. AI agents can sometimes get all the way to “buy,” then stall, forced to hand control back to a human to complete the transaction. That breaks the promise of end-to-end automation and introduces the very friction agents were designed to remove. It’s not only frustrating but it slows commerce and, of course, revenue.

The core issue is AI agents aren’t just faster humans. They're a new customer type entirely.

Let the Flow Continue to Payment

Agents don’t “check out” the way people do. They don’t fill out forms or click buttons on user interfaces designed for human interaction. By their very nature, they operate and communicate using APIs and AI-native interfaces such as Model Context Protocol (MCP). Using this communication model, agents navigate inside workflows such as email threads, Slack conversations, procurement systems, calendaring, and simply executing tasks continuously and autonomously. When retailers treat agents like human shoppers, their workflows break and the shopping cart is left unpaid until a human logs in to complete the payment process.

More importantly, retail platforms today have no way to recognize or onboard agents as legitimate customers. Merchants can’t provision an agent as a discrete buyer tied to a real human or organization. They can’t define permissions, spending limits or accountability rules in a way that allows agents to transact safely. And they can’t accept payments that are native to agents — payments that are authorized programmatically, traceable to a human owner, and auditable after the fact.

This Gap Creates a Massive Growth Constraint

The next wave of commerce won’t be driven by better product recommendations or flashier user experience. It will be driven by eliminating friction between intent and transaction and, of course, inside the workflows where agents already operate — at the very moment the transaction needs to happen. If an agent identifies the right product during a Slack conversation between business users or a procurement flow to keep agents working, commerce should happen there. No redirects. No reauthentication. No manual approvals that grind work and business to a halt.

Retailers that fail to adapt will see agent-driven demand leak elsewhere. Agents will naturally favor merchants that can transact cleanly, reliably and programmatically. Those that cling to browser-based, human-only checkout motions will be invisible to this rapidly emerging customer segment.

Winners Will Build Agent-First Commerce Paths

What this means is exposing API-driven purchasing instead of forcing agents through web forms. It means supporting embedded payments that allow agents to transact within existing systems of work. It means improving today’s identity logic to bind agents to real humans or businesses, with clear guardrails around permissions, limits and accountability. And it means rethinking fraud, compliance, and trust models for a world where software — not people — initiates the purchase.

What’s more, AI agents should also be able to use different payment types, essentially the widest range of monetization models. This includes one-time payment, recurring, micropayments or hybrid. If the system is built to serve different buyer needs and various payment types, the buyer has zero friction and the merchant wins.

We’re Not Replacing Humans. We’re Enabling Agents to Execute End-to-End Under Policy Control

When agents can complete purchases autonomously, humans move up the value chain. This translates to setting strategy, defining constraints, and making higher-order decisions instead of clicking through checkout screens. Retailers benefit from faster cycles, higher conversion, and new sources of demand that simply don’t exist in today’s manual flows. In fact, it’s a competitive advantage to provide this level of flexible, modern way of doing business.

AI agents are already your customers. The important question now is whether your commerce stack is ready to serve them or whether they’ll quietly take their business to someone that is.

Jim Nguyen is the co-founder and CEO of InFlow, an AI-native payments platform built for agentic commerce.