A unified product data layer for the agentic commerce era
AI agents are changing how people shop—not by adding another app, but by becoming the interface to every product online. They find what consumers want, compare options, and decide where to buy in seconds. To make that possible, they need access to clean, structured, and trusted product data at scale.
Klarna’s
defines an open standard for how product and offer data is structured, identified, and exchanged across the web. It specifies how merchants, aggregators, and AI systems can communicate using a shared schema, making it easier for agents to understand what products exist, where they’re sold, and at what price or availability.Built on Klarna’s global product graph and data infrastructure, our protocol complements the Agentic Commerce Protocol introduced by Stripe and OpenAI and provides instant access to 100M+ products and 400M+ prices—powering the discovery and product data layer that precedes agentic checkout.
Anyone can implement the protocol to expose product data in a standardized way, enabling interoperability between merchants, aggregators, and AI systems.
What it enables
Klarna’s Agentic Product Protocol introduces a consistent framework for describing products, categories, and offers, abstracting away differences between merchant systems.
It enables agents to:
Search for and compare the same product across multiple merchants seamlessly
Maintain persistent product references over time (e.g., for “save for later” or price alerts)
Reason over structured attributes like brand, model, and condition
Interoperate with other systems through open, documented formats
In short, our Agentic Product Protocol turns the world’s fragmented product data into a single, structured source of truth for AI.
A unified taxonomy for discovery
The protocol introduces a multi-level taxonomy that organizes products into hierarchical categories, from broad types down to specific attributes. This structure allows agents to drill down intelligently: from “electronics” to “smartphones” to “Android phones.”
Each product includes core information independent of any single merchant, such as images, descriptions, brand, and key attributes. By aggregating this information under a shared and consistent product ID, the protocol creates a unified representation of a product across all sellers, rather than treating each merchant listing as separate.
Multiple offers (price, stock, and merchant data) can then be linked to the same product ID. This means agents can instantly compare options across merchants without parsing hundreds of inconsistent listings.
Why this matters for AI discovery
Search in the era of AI becomes more accurate, faster, and more useful when agents can query structured product entities instead of text-based listings.
With product normalization, AI systems can:
Return relevant results even when merchants use different naming or categorization conventions
Group and compare identical products across multiple sellers
Retrieve the same product later (for features like “save for later” or price tracking)
How Klarna’s Agentic Product Protocol works
Shopper:
The shopper asks an AI agent a shopping-related question—for example, “find me the best wireless headphones in pink.” The agent uses product data exposed through the Agentic Product Protocol to search across multiple merchants, compare options, and display relevant results with pricing and availability.
AI agent:
The agent interprets the shopper’s intent, queries the Agentic Product Protocol data layer (directly or via the Agentic Product Protocol API), and retrieves structured product entities with multiple offers. It can reason over attributes like brand, model, or specifications to present the most relevant results. When the shopper shows interest, the agent can hand off to a checkout provider to complete the purchase.
Merchant:
Merchants can share their existing product feeds with Klarna to be integrated directly into the Agentic Product Protocol API, or choose to expose their product data through the Agentic Product Protocol standard themselves for interoperability. This makes their products discoverable by any AI system connected to the Agentic Product Protocol, without needing custom integrations with every agent or platform.
Data provider / Aggregator:
Aggregators and platforms can ingest merchant feeds, normalize them using the Agentic Product Protocol schema, and expose them as standardized endpoints. This ensures agents have access to continuously refreshed, multi-merchant data in a consistent format that’s easy to query and scale globally.
The below shows an example of how the Agentic Product Protocol can easily be implemented using the Agentic Product Protocol API.
Built for openness and scale
The Agentic Product Protocol is designed to be independent and accessible to any developer building agentic or commerce-driven experiences. It’s built around Klarna’s principles of transparency, compatibility, and scalability, ensuring that any participant — from a single retailer to a global marketplace — can connect using familiar data formats.
Access 100M+ products and 400M+ prices instantly with the Agentic Product Protocol API
The Agentic Product Protocol API is Klarna’s hosted implementation of the protocol—a ready-to-use, data-rich API that gives developers and AI systems instant access to structured, continuously refreshed product data.
Instead of building and maintaining individual merchant integrations, connect once to access 100M+ products and 400M+ prices across the US and 11 European markets — powered by over 1 billion prices processed every day.
No scraping. No patchwork integrations. Just clean, normalized product data through a single API.
API benefits:
Instant access to the protocol’s data layer – pre-filled with normalized product and offer data from hundreds of merchants.
Unified taxonomy – A curated, cross-merchant category structure for consistent product classification and easier multi-merchant search.
Flexible ingestion formats – Support for multiple feed standards, including Google Merchant, Facebook Catalog, Shopify APIs, and generic CSV or JSON formats.
How Klarna’s Agentic Product Protocol API compares
Independent Multi-Merchant Coverage
Our platform aggregates data directly from a wide network of merchants across both the US and EU. Unlike Shopify, which is limited to its own merchant base, or Google Shopping, which mirrors Google’s listings, we maintain independent coverage of the open retail landscape, giving AI systems visibility into where and how products are sold across markets.
Product Deduplication / Normalization
Our platform performs true normalization, resolving and merging identical products across merchants into stable product IDs with consistent attributes.
This structured foundation enables accurate product comparison and reasoning.
Multiple Offers per Product Entity
Where Shopify exposes one listing per merchant or snapshot, our system links multiple merchant offers to each normalized product entity. This creates a single, unified view of all available prices and options for that product.
Continuously Updated Offers
We track prices, stock status, and availability through a mix of real-time and scheduled updates, ensuring results stay current and trustworthy. Unlike batch-based providers such as Dataweave, Klarna Agentic Product Protocol update pipeline keeps information continuously in sync across markets.
AI / Agentic Commerce Ready
Our API outputs structured, machine-interpretable data ready for LLMs and autonomous agents. Competitors like Dataweave and Shopify focus on analytics or merchant integrations, not AI consumption.
API-First Integration Model
Our platform was designed as a developer-first API with modern endpoints. Google Shopping offers simple REST APIs; Dataweave leans on enterprise feed delivery. We combine developer simplicity with enterprise scalability, making integration seamless for AI builders.
LLM-Optimized Responses (JSON / Embeddings)
Our responses are structured and vector-ready, enabling LLMs to reason over product data and integrate it into retrieval-augmented or agentic workflows.
Others return unstructured text, requiring additional parsing or fine-tuning.
Global Market Coverage
We maintain consistent category, taxonomy, and merchant mapping across both regions, ensuring a uniform data model that scales internationally and which is continuously growing. By contrast, Shopify’s data is limited to its own merchant base, Google Shopping mirrors Google’s regional search results, and neither provides global, cross-merchant structured product coverage.
Our coverage includes:
US market: Broad integration with leading online retailers, marketplaces, and direct merchant feeds
EU market: Expansive coverage in the United Kingdom, Germany, France, Italy, Spain, Netherlands, Sweden, Denmark, Norway, Finland, Ireland, and Austria.
Unified schema: A single, normalized data model ensures product and offer consistency across markets
We’re continuously expanding to new markets and onboarding additional merchants to broaden access to global product data.
How to get started
Developers
Access to the API is available for approved partners and developers.
Merchants
Are you a merchant who wants your products to be part of Klarna Agentic Product Protocol? Join the network powering the next generation of AI-driven commerce