All Posts

BNPL and Agent Commerce: How AI Agents Handle Buy Now Pay Later at Checkout

A guide for developers building AI agents that need to handle BNPL payment options at checkout, covering the technical challenges and how UCP simplifies payment method selection.

September 21, 2026UCPList Team
BNPLbuy now pay laterUCPagent commerceAI shopping agents

Buy now pay later has gone from a niche checkout option to a mainstream payment method. Klarna, Afterpay, Affirm, Sezzle, Paidy, and dozens of regional competitors now appear at checkout across millions of merchants. For AI agents handling purchases on behalf of consumers, this creates a practical problem: how do you select the right payment method when the consumer's preferred option is an installment plan rather than a credit card?

Why BNPL Matters for Agent Commerce

Consumer preferences around payment are not uniform. A consumer who uses Afterpay for every fashion purchase, or Paidy for every purchase on a Japanese retailer, expects an agent acting on their behalf to respect that preference. An agent that defaults to credit card at checkout and ignores available BNPL options is making a financial decision for the consumer without their input.

BNPL also introduces an approval step. Unlike stored credit cards, BNPL providers run a soft credit check at the time of the transaction. The consumer gets approved (or declined) at checkout. This is different from the predictable success of charging a stored card. An agent needs to handle approval, decline, and fallback states gracefully.

There is also a geography problem. BNPL market share varies significantly by country. In Australia and New Zealand, Afterpay and Zip command large shares of checkout volume. In Japan, Paidy is the dominant deferred payment option. In the US, Affirm has a stronger position in higher-value purchases while Afterpay and Sezzle lead in fashion and lifestyle. A well-built agent should know which BNPL providers are likely available in a given market.

The Technical Challenge at Checkout

BNPL payment selection happens at checkout, not at the payment processing layer. The challenge for agents is that BNPL options are surfaced differently depending on the merchant's platform and the BNPL provider's integration.

On Shopify merchants, BNPL options appear as payment method widgets in the checkout. Afterpay, Klarna, and Sezzle have native Shopify integrations that render their UI in the payment step. An agent interacting with a Shopify checkout needs to identify that a BNPL option is available and select it, then handle the redirect or iframe flow that the BNPL provider uses to complete the approval.

On merchants using custom checkout implementations, BNPL availability is less predictable. Some merchants expose it as a standard payment method option. Others require a separate flow triggered by clicking a BNPL widget. The rendering and interaction model varies by provider.

Payment tokens are the other complexity. BNPL providers do not issue stored payment tokens in the same way that credit card processors do. The approval is transaction-specific. An agent cannot store a Klarna token and reuse it across purchases the way it would with a tokenized credit card.

How UCP Simplifies BNPL Selection

UCP standardizes the payment method negotiation layer at checkout. A UCP-compliant merchant endpoint exposes available payment methods as part of the checkout capability manifest. This means an agent can query the endpoint, retrieve the list of accepted payment methods including BNPL options, and select accordingly.

Rather than scraping checkout HTML to detect Afterpay widgets or parsing merchant-specific checkout APIs to find BNPL options, the agent reads a structured capability response:

{
  "paymentMethods": [
    { "type": "card", "networks": ["visa", "mastercard"] },
    { "type": "bnpl", "provider": "afterpay", "markets": ["US", "AU", "CA"] },
    { "type": "bnpl", "provider": "klarna", "markets": ["US", "DE", "SE"] }
  ]
}

The agent matches consumer payment preferences against the available methods, selects the appropriate BNPL provider, and proceeds through the UCP checkout flow. The BNPL approval step is handled within the UCP payment token exchange, which abstracts the provider-specific approval redirect into a structured response.

Practical Implementation Notes

When building agent payment logic for BNPL, handle these cases:

Approval decline. BNPL providers decline transactions. Your agent needs a fallback strategy: try a different BNPL provider if one is available, or fall back to the consumer's stored card. Log the decline for the consumer to review.

Market availability. Before selecting a BNPL provider, check that it operates in the consumer's market. Sezzle is North America only. Paidy is Japan only. Laybuy operates in UK, NZ, and AU. An agent selecting a BNPL provider unavailable in the consumer's region will generate an error at checkout.

Purchase limits. BNPL providers have per-transaction limits. Afterpay's limit varies by consumer history. Affirm handles higher-value purchases but has a minimum. Check limits before selecting BNPL for large purchases.

Consumer consent. BNPL creates a financial obligation. Make sure your agent surfaces the installment schedule to the consumer before confirming the purchase. An agent that silently selects BNPL without disclosure creates a bad experience.

What to Watch

Klarna and Afterpay have both been active in the UCP ecosystem conversations. As the protocol matures, expect BNPL providers to publish UCP endpoints that standardize the approval and token exchange flow. This will make BNPL selection significantly more reliable for agents than the current scraping and provider-specific SDK approach.

For developers building agents that handle purchases across multiple markets, start by mapping which BNPL providers are used in your target markets, then build preference logic that matches consumer settings to available options at checkout. The UCPList directory tracks BNPL providers in the payment-handlers category so you can see which ones are active in the ecosystem.

Read next