Klarna Wants AI Agents to Shop for You. Here’s How.

According to PYMNTS.com, Klarna launched its new Agentic Product Protocol (APP) specification and API on Monday, December 15. The open standard provides a live, structured feed of a staggering 100 million products and 400 million prices, all standardized across 12 different markets. The protocol, announced by Klarna Chief Commercial Officer David Sykes, is designed to establish a common language for AI systems, merchants, and platforms to exchange product data. For merchants, the API makes products discoverable to any AI agent that supports the protocol and supports feed formats from Google Merchant, Shopify, Facebook Catalog, and CSV/JSON. This follows Klarna’s October announcement of an expanded partnership with Google to support the Agent Payments Protocol (AP2), an open standard for secure, AI-driven payments. The core idea, as Sykes put it, is that “before agents can buy, they need to know what exists.”

The New Gatekeepers

Here’s the thing: Klarna isn’t just building a better product feed. They’re trying to build the plumbing for a whole new era of commerce. For decades, discovery was about SEO and search ads. Then it shifted to social media feeds and influencer hauls. The next frontier, according to basically everyone in tech, is AI agents that shop for you. But those agents need structured, reliable, and comprehensive data to work. If your product listing has poor images or incomplete attributes, you might as well be invisible. You’re not just failing to rank on Google; you’re being excluded from the AI agent’s response entirely. Klarna is positioning itself to be the source of truth that prevents that exclusion. It’s a smart, defensive play for a company whose core business is payments at the point of purchase. If you own the discovery layer, you’re in a powerful position to influence the transaction that follows.

Beyond Keywords

This shift changes everything about how we think about online shopping. It’s not about typing “blue running shoes” into a search bar anymore. It’s about telling an agent, “Find me a durable pair of trail runners for under $120 that are good for wide feet, and see if any local stores have them in stock for pickup today.” That’s a contextual, conversational query. To answer it, an AI needs to understand product specs, inventory levels, pricing across retailers, and local logistics—all in real time. Klarna’s protocol is an attempt to standardize at least the product data part of that massive equation. But it raises a big question: in a world of AI agents, who controls the customer relationship? Is it the merchant, the AI platform (like Google or a future OpenAI shopping agent), or the infrastructure provider like Klarna in the middle? The battle isn’t for the checkout button anymore; it’s for the conversation that happens before the cart is even created.

The Race for Standardization

Klarna’s move with Google on payments (AP2) and now this product protocol shows they’re thinking in layers. They want to define the foundational standards for agentic commerce. It’s a classic tech strategy: commoditize the complement. If they can make product discovery and AI payments frictionless through open protocols, their own payment service becomes the obvious, integrated choice. But they’re not the only ones thinking this way. You can bet Amazon, Shopify, and every major platform is working on their own agent strategies. The risk for merchants will be fragmentation—having to support a dozen different “open” standards. Klarna’s bet is that by releasing their spec openly and leveraging their existing merchant network, they can achieve critical mass first. It’s an ambitious land grab. And if it works, Klarna evolves from a “buy now, pay later” fintech into the essential data backbone for the next generation of automated shopping. That’s a much bigger story.