Beyond the purchase: How Checkout.com is thinking about agentic commerce

Artificial intelligence is reshaping commerce. For decades, ecommerce revolved around websites, apps, and digital storefronts that required consumers to manually browse, compare, and buy. But a new channel is emerging – one where AI-powered agents shop on behalf of consumers. We call this agentic commerce.

At Checkout.com, we believe agentic commerce represents a fundamental shift: from transactions being initiated by people to being delegated to trusted AI assistants. These AI agents are more than passive tools – they act with intent, making decisions, managing subscriptions, discovering products, and negotiating the buying journey on behalf of their users.

But for agentic commerce to scale, it requires infrastructure. Payments need to be secure, trusted, and transparent; merchants need control, visibility, and influence; and AI platforms need access to verified credentials and policies. This is where Checkout.com can help.

Building blocks of secure agentic payments

We have the privilege of being at the intersection between merchants and AI platforms. Every day, we help some of the world’s largest brands accept payments through APIs that are already powering digital commerce at scale.

Now, we are extending that role into the agentic era – building the bridge that allows merchants to safely transact with AI agents, while ensuring agents act responsibly and intelligently on behalf of consumers.

Today, agents are visiting merchant websites unseen – browsing, scraping, and sometimes purchasing without the merchant’s awareness. Our thinking is to move beyond simply processing agent payments. The standard must evolve from enabling transactions to enabling collaborative, intelligent commerce. 

Tomorrow, we see a world where merchants and agents interact transparently, exchanging trusted signals that enhance personalization, security, and fulfillment.

Together with Visa and Mastercard, we’re laying the foundations for an agent-powered economy. The building blocks include:

1. Agent onboarding & identity

Agents must be identifiable participants in commerce. Just as we have Know Your Customer and Know Your Merchant standards, we’re developing Know Your Agent (KYA) frameworks. AI agents receive a digital fingerprint, ensuring merchants, issuers, and schemes all know when a transaction is being initiated by an agent.

2. Agent-specific credentials

Agents need their own secure credentials to transact. Relying on consumer-issued cards or disposable virtual cards is neither scalable nor safe. That’s why Checkout.com is developing an Agentic Credentials Framework – a flexible system that equips AI agents with the right type of credential for the right context.

• Agent-specific network tokens: Tokens issued with unique identifiers and rules that make it clear when an AI agent, rather than a human, is initiating a transaction.
• Virtual cards: Still useful in certain contexts, but integrated as part of a managed framework rather than a patchwork workaround.
• Wallet-based credentials: Allowing agents to transact via wallets, with the same security and transparency as human-led commerce.
• Crypto or exchange-linked credentials: Enabling future agent transactions in alternative payment methods, anchored in the same trust standards.

This framework ensures that agents are provisioned with the safest, most efficient credential for each transaction type – reducing costs, improving performance, and keeping merchants and issuers in control.

3. Authentication

Agents need a secure way to authenticate consumers on their behalf. Checkout.com provides this through a biometric-first, phishing-resistant framework built on modern standards.

• Smooth experience: Once a consumer registers with Checkout.com authentication, they can use it across any participating merchant, ensuring a consistent and simple biometric flow.
• Trusted by design: Credentials never leave the consumer’s device, meeting the highest security and regulatory standards.
• Agent-safe: Agents never hold sensitive credentials – only a scoped, verifiable proof is passed into the transaction.

4. Personalization signals

Agentic commerce isn’t only about secure payments – it’s about making AI-powered shopping smarter. With consumer consent, Checkout.com can enrich AI platforms with context derived from transaction networks (ours, Visa, and Mastercard).

For example:

• A consumer provides their card details to an AI platform to book a hotel.
• Checkout.com checks these details across our network and scheme partners.
• We return insights (with consumer consent) such as: this consumer frequently shops at Starbucks.
• The AI platform can then refine results and suggest hotels close to Starbucks, creating hyper-personalized recommendations that reflect real-world behavior.

This closes the gap between financial data and agentic decision-making – enabling AI platforms to act more like true personal assistants.

5. Consent & control management

For agentic commerce to work at scale, consumers must define the boundaries of what their AI agents can do on their behalf. This goes beyond simple authentication and into transaction-level intent.

Checkout.com enables AI platforms to capture and enforce this consent, such as:

• Spending limits – e.g., “do not spend more than $300.”
  Conditional purchases – e.g., “buy the Sony headphones only if the price drops to $300.”
• Frequency and timing – e.g., “book flights once per quarter” or “only make purchases after payday.”

Capturing and recording this consent is critical for two reasons:

1. Operational trust – AI platforms can act confidently within consumer-defined boundaries.
2. Downstream assurance – if a consumer disputes a transaction (e.g., claiming fraud), Checkout.com can provide issuers with the original consent record. This strengthens chargeback and dispute programs by showing whether the transaction is aligned with consumer-approved parameters.

In short, consent and control management ensures that agent-led transactions are traceable, defensible, and trusted across the entire ecosystem – from consumer to agent, merchant, and issuer.

Beyond payments: Smarter product discovery

While payments are the foundation, the opportunity in agentic commerce extends into product discovery and decision-making.

Today, many AI platforms rely on two fragile tools: virtual cards and web crawling.

• Virtual cards: AI platforms sometimes issue disposable cards to simulate consumer payments. But this is an expensive and inefficient approach – generating cards at scale adds significant cost, complexity, and compliance burden (e.g., PCI DSS). It also isn’t the safest option: schemes and regulators are increasingly scrutinizing this model, and many merchants actively restrict virtual cards due to higher levels of fraud. The result is more friction, not less.
  
  
• Crawling online stores: Agents scrape websites to understand pricing, availability, and product details. This is computationally heavy, prone to errors, and not always safe. It degrades the consumer experience because of delays between when a consumer confirms intent and when the agent actually executes a purchase. In some cases, the information is already outdated by the time the agent acts.

Both approaches are costly, brittle, and far from scalable.

The next evolution in commerce isn't just about faster payments; it's about smarter data. Imagine a future built on a 'network of networks,' enabling AI agents to access real-time inventory from any merchant. This concept of Inventory Reach would replace inefficient web scraping with direct queries, leading to a new era of speed, accuracy, and relevance in consumer AI.

The potential of this model requires two-way information flows between merchants and AI platforms, agents can move from guesswork to real-time, contextualized intelligence. This shift not only makes shopping more seamless, but also ensures merchants stay in control, consumers stay protected, and AI platforms can operate at scale without regulatory risk.

This two-way information exchange creates a step change:

Merchants → Checkout.com → Agents

• Real-time pricing and stock levels
• Personalized fulfillment options (e.g., loyalty-based delivery upgrades)
• Clear return policies and purchase conditions‍

‍Agents → Checkout.com → Merchants

• Real-time intent signals (“shopping for hiking boots in Colorado”)
• Consumer preferences and context (loyalty status, preferred payment method)
• Linked browsing sessions tied to customer accounts

This creates a commerce loop where AI-powered decisions are both personalized and transparent.

Guiding agents inside the store

One of the most exciting frontiers is allowing merchants to set guardrails and guidance for agents.

Imagine an agent enters a digital storefront:

• The merchant signals navigation cues (“highlight sustainable fashion options for this customer”).
• The checkout process requests additional ID verification for high-value items.
• Premium customers are offered free express delivery – communicated directly to the agent.

This is more than transactions. It's a digital collaboration between merchants and AI agents, orchestrated through APIs provided by Checkout.com.

The roadmap ahead

Agentic commerce is in its infancy, but we see a clear evolution:

Checkout.com is committed to building this future responsibly. By partnering with schemes, merchants, and AI platforms, we’re establishing the standards and infrastructure that make agentic commerce safe, scalable, and collaborative.

Closing thought

Agentic commerce will redefine the relationship between consumers, merchants, and payments. It is not just a new sales channel – it is a new operating model for digital commerce.

At Checkout.com, we are proud to help shape this transformation:

• Securing payments through agent onboarding, credentials, and authentication
• Enabling trust with schemes and issuers through transparent standards
• Empowering merchants with visibility, guardrails, and personalization
• Enhancing discovery by bridging agents and merchants with real-time information

The future of commerce is one where AI doesn’t just transact – it collaborates. And Checkout.com is building the rails to make that possible.