News

Growth

Success stories

Papers

Webinars & events

News

Growth

Success stories

Papers

Webinars & events

News

Growth

Success stories

Papers

Webinars & events

News

Growth

Success stories

Papers

Webinars & events

November 2, 2026

Agentic commerce: your questions answered

Agentic commerce: your questions answered

Agentic commerce: your questions answered

Agentic commerce: your questions answered

Our payment infrastructure is ready for the future of AI sales. Now, read our expert guide to prepare for the agentic commerce era.

Our payment infrastructure is ready for the future of AI sales. Now, read our expert guide to prepare for the agentic commerce era.

Agentic commerce is no longer just a concept. Early commerce experiences are live in the US, and the protocols and payments infrastructure that could bring them to Europe are developing quickly.

For your business, agentic commerce changes how customers discover, compare, and buy – especially as more of that journey moves through an AI agent instead of a browser.

In this guide, we answer the most pressing questions on competing protocols, payment security, and what this channel really means for you.

Looking for a more basic definition? Go to our article ‘What is agentic commerce?’

Agentic commerce is no longer just a concept. Early commerce experiences are live in the US, and the protocols and payments infrastructure that could bring them to Europe are developing quickly.

For your business, agentic commerce changes how customers discover, compare, and buy – especially as more of that journey moves through an AI agent instead of a browser.

In this guide, we answer the most pressing questions on competing protocols, payment security, and what this channel really means for you.

Looking for a more basic definition? Go to our article ‘What is agentic commerce?’

Agentic commerce is no longer just a concept. Early commerce experiences are live in the US, and the protocols and payments infrastructure that could bring them to Europe are developing quickly.

For your business, agentic commerce changes how customers discover, compare, and buy – especially as more of that journey moves through an AI agent instead of a browser.

In this guide, we answer the most pressing questions on competing protocols, payment security, and what this channel really means for you.

Looking for a more basic definition? Go to our article ‘What is agentic commerce?’

Agentic commerce is no longer just a concept. Early commerce experiences are live in the US, and the protocols and payments infrastructure that could bring them to Europe are developing quickly.

For your business, agentic commerce changes how customers discover, compare, and buy – especially as more of that journey moves through an AI agent instead of a browser.

In this guide, we answer the most pressing questions on competing protocols, payment security, and what this channel really means for you.

Looking for a more basic definition? Go to our article ‘What is agentic commerce?’

What this means for your business

Agentic commerce could change how customers find, compare, and buy from you. Instead of only optimising for human shoppers, you may also need to make your products, policies, and checkout available to AI agents.

The immediate priorities are clear: make key business data machine-readable, keep your ecommerce setup healthy, understand how AI tools describe your brand, and work with a payment partner that can support emerging agentic standards.

Agentic commerce could change how customers find, compare, and buy from you. Instead of only optimising for human shoppers, you may also need to make your products, policies, and checkout available to AI agents.

The immediate priorities are clear: make key business data machine-readable, keep your ecommerce setup healthy, understand how AI tools describe your brand, and work with a payment partner that can support emerging agentic standards.

Agentic commerce could change how customers find, compare, and buy from you. Instead of only optimising for human shoppers, you may also need to make your products, policies, and checkout available to AI agents.

The immediate priorities are clear: make key business data machine-readable, keep your ecommerce setup healthy, understand how AI tools describe your brand, and work with a payment partner that can support emerging agentic standards.

Agentic commerce could change how customers find, compare, and buy from you. Instead of only optimising for human shoppers, you may also need to make your products, policies, and checkout available to AI agents.

The immediate priorities are clear: make key business data machine-readable, keep your ecommerce setup healthy, understand how AI tools describe your brand, and work with a payment partner that can support emerging agentic standards.

What does the agentic ecosystem look like?

Agentic commerce doesn’t happen in a single place. It’s a new, complex ecosystem that operates through a chain of AI systems interacting with your business systems.

There are three key layers:

The consumer agent (the buyer): This is the AI assistant the customer interacts with, such as ChatGPT, Gemini, or another AI-powered shopping experience. Its job is to understand the customer’s complex request (e.g., "Find me a sustainable, size 10, waterproof rain jacket for under €150") and turn it into actionable steps. Customers use agents in two ways. In assisted (co-pilot) mode, the agent researches and recommends, and the customer buys either on the business's site or inside the chat. In delegated mode, the agent buys on the customer’s behalf within rules the customer sets.
The protocol (the translator): The protocol is the set of rules that allows a consumer agent to securely and programmatically communicate with a business’s systems. It standardises how product information is requested and how a secure payment is initiated.
The business system (the seller): This is your API-first architecture, which exposes machine-readable data on your products, stock levels, pricing, and policies. Most brands manage their catalogue and inventory through an ecommerce platform such as WooCommerce, Magento, Shopify, or one of the many local platforms in Europe. Your system must be ready to receive a request from the agent (via the protocol) and process a secure, tokenised payment through a payment partner like Mollie.

Post-purchase support is part of the picture too. Agents can answer questions about delivery, returns, and chargebacks, so your service after the sale matters more than ever.

Mollie sits at the intersection of the protocol and the payment processing layer, ensuring that when the agent makes a purchase, your business gets paid securely and compliantly.

Agentic commerce doesn’t happen in a single place. It’s a new, complex ecosystem that operates through a chain of AI systems interacting with your business systems.

There are three key layers:

The consumer agent (the buyer): This is the AI assistant the customer interacts with, such as ChatGPT, Gemini, or another AI-powered shopping experience. Its job is to understand the customer’s complex request (e.g., "Find me a sustainable, size 10, waterproof rain jacket for under €150") and turn it into actionable steps. Customers use agents in two ways. In assisted (co-pilot) mode, the agent researches and recommends, and the customer buys either on the business's site or inside the chat. In delegated mode, the agent buys on the customer’s behalf within rules the customer sets.
The protocol (the translator): The protocol is the set of rules that allows a consumer agent to securely and programmatically communicate with a business’s systems. It standardises how product information is requested and how a secure payment is initiated.
The business system (the seller): This is your API-first architecture, which exposes machine-readable data on your products, stock levels, pricing, and policies. Most brands manage their catalogue and inventory through an ecommerce platform such as WooCommerce, Magento, Shopify, or one of the many local platforms in Europe. Your system must be ready to receive a request from the agent (via the protocol) and process a secure, tokenised payment through a payment partner like Mollie.

Post-purchase support is part of the picture too. Agents can answer questions about delivery, returns, and chargebacks, so your service after the sale matters more than ever.

Mollie sits at the intersection of the protocol and the payment processing layer, ensuring that when the agent makes a purchase, your business gets paid securely and compliantly.

Agentic commerce doesn’t happen in a single place. It’s a new, complex ecosystem that operates through a chain of AI systems interacting with your business systems.

There are three key layers:

The consumer agent (the buyer): This is the AI assistant the customer interacts with, such as ChatGPT, Gemini, or another AI-powered shopping experience. Its job is to understand the customer’s complex request (e.g., "Find me a sustainable, size 10, waterproof rain jacket for under €150") and turn it into actionable steps. Customers use agents in two ways. In assisted (co-pilot) mode, the agent researches and recommends, and the customer buys either on the business's site or inside the chat. In delegated mode, the agent buys on the customer’s behalf within rules the customer sets.
The protocol (the translator): The protocol is the set of rules that allows a consumer agent to securely and programmatically communicate with a business’s systems. It standardises how product information is requested and how a secure payment is initiated.
The business system (the seller): This is your API-first architecture, which exposes machine-readable data on your products, stock levels, pricing, and policies. Most brands manage their catalogue and inventory through an ecommerce platform such as WooCommerce, Magento, Shopify, or one of the many local platforms in Europe. Your system must be ready to receive a request from the agent (via the protocol) and process a secure, tokenised payment through a payment partner like Mollie.

Post-purchase support is part of the picture too. Agents can answer questions about delivery, returns, and chargebacks, so your service after the sale matters more than ever.

Mollie sits at the intersection of the protocol and the payment processing layer, ensuring that when the agent makes a purchase, your business gets paid securely and compliantly.

Agentic commerce doesn’t happen in a single place. It’s a new, complex ecosystem that operates through a chain of AI systems interacting with your business systems.

There are three key layers:

The consumer agent (the buyer): This is the AI assistant the customer interacts with, such as ChatGPT, Gemini, or another AI-powered shopping experience. Its job is to understand the customer’s complex request (e.g., "Find me a sustainable, size 10, waterproof rain jacket for under €150") and turn it into actionable steps. Customers use agents in two ways. In assisted (co-pilot) mode, the agent researches and recommends, and the customer buys either on the business's site or inside the chat. In delegated mode, the agent buys on the customer’s behalf within rules the customer sets.
The protocol (the translator): The protocol is the set of rules that allows a consumer agent to securely and programmatically communicate with a business’s systems. It standardises how product information is requested and how a secure payment is initiated.
The business system (the seller): This is your API-first architecture, which exposes machine-readable data on your products, stock levels, pricing, and policies. Most brands manage their catalogue and inventory through an ecommerce platform such as WooCommerce, Magento, Shopify, or one of the many local platforms in Europe. Your system must be ready to receive a request from the agent (via the protocol) and process a secure, tokenised payment through a payment partner like Mollie.

Post-purchase support is part of the picture too. Agents can answer questions about delivery, returns, and chargebacks, so your service after the sale matters more than ever.

Mollie sits at the intersection of the protocol and the payment processing layer, ensuring that when the agent makes a purchase, your business gets paid securely and compliantly.

banner payments report 2026

banner payments report 2026

The Payments Report 2026

Get the operational playbook to automate complexity and turn transaction data into revenue.

Download the report

Download the report

What are the competing agentic commerce protocols?

For payments to happen securely inside an AI conversation, everyone needs to follow a standard set of rules. This is where protocols come in. Currently, the main players are ACP, AP2, and Google’s Universal Commerce Protocol (UCP).

These protocols are still evolving, and they solve slightly different parts of the agentic commerce puzzle. Here’s how they compare.

What is ACP (Agentic Commerce Protocol)?

The Agentic Commerce Protocol (ACP) is the standard created by OpenAI and Stripe.

Status: ACP launched in September 2025 through ChatGPT's Instant Checkout, starting with Etsy and Shopify sellers in the US. Take-up stayed low, and OpenAI retired Instant Checkout in March 2026. The protocol itself remains active, but OpenAI’s focus has shifted from a single in-chat buy button towards business-built apps and merchant-controlled checkout experiences inside ChatGPT.
Focus: ACP enables human-present (assisted shopping) flows, allowing a customer to complete a purchase securely within their conversation with an AI agent. It also includes the APIs that let brands expose their inventory and finalise checkouts.
Our strategy: We built Mollie's compatibility with ACP's Delegated Payment API early, while the protocol was still centred on in-chat checkout. We're keeping that infrastructure current as OpenAI's own approach evolves, so you're ready to get paid through whichever ACP-powered surface – in-chat or a merchant app – reaches your customers.

What is AP2 (Agent Payments Protocol 2)?

The Agent Payments Protocol 2 (AP2) is an open-source framework introduced by Google Cloud, Coinbase, and over 60 industry partners to enable secure, verifiable, autonomous payments for AI agents.

AP2 extends the open-source Agent2Agent (A2A) protocol to handle payments, and UCP in turn uses AP2 as its payment layer.

Status: AP2 has released v0.2 and its development has moved under the FIDO Alliance. It’s designed for production use, but widespread live consumer checkouts are still emerging.
Focus: AP2 was designed to support human-absent (delegated shopping) flows from day one. This means the AI can act fully on the customer’s behalf to complete a purchase based on a pre-approved mandate (e.g., “Buy me new coffee beans when I run out”). It also supports human-present (assisted) shopping.
Our strategy: We’re an active partner in the AP2 consortium, helping shape a standard that serves long-term European needs for flexibility and security. We want all payment methods, not only cards, to work in an agentic checkout.

What is UCP (Universal Commerce Protocol)?

The Universal Commerce Protocol (UCP) is an open-source standard co-developed by Google and industry partners to enable agentic commerce across AI platforms, web apps, and business backends.

Status: UCP is live in the US on Google’s AI surfaces, AI Mode in Google Search and Gemini. It’s still an evolving standard, and not every feature in the specification is available on those surfaces yet. Merchants must join a waitlist and be approved by Google before they can go live.
Focus: UCP offers two integration paths. In the native checkout path, customers complete the purchase on Google’s AI surfaces using a Google Pay payment handler, and the business sends order updates after purchase. In the cart path, customers move from Google’s surfaces to the business's website to complete the purchase.
Our strategy: We support all the payment protocols that are made available, and we’re following the industry’s move toward UCP. Our aim is to make sure the standard works for European merchants and consumers, including local payment methods.

ACP, AP2, and UCP: Key differences

Feature	ACP (Agentic Commerce Protocol)	AP2 (Agent Payments Protocol 2)	UCP (Universal Commerce Protocol)
Status	Launched 2025, now powers business apps in ChatGPT	Released as v0.2, now under FIDO Alliance governance; built for production, but live consumer checkouts are still emerging	Live on Google AI surfaces (AI Mode in Search and Gemini), with business approval required
Primary goal	Primarily designed for human-present (assisted) flows	Designed for both human-absent (delegated) and human-present (assisted) flows	Standardises the full purchase process: cart, checkout, and payment
Analogy	A personal shopper helping a customer find and buy items	A personal shopper completes a purchase independently	A shop assistant who handles the whole purchase in one shared format
Key players	OpenAI / Stripe / PayPal (Mollie built compatibility with ACP's Delegated Payment API)	Google / Partners (including Mollie)	Google

For payments to happen securely inside an AI conversation, everyone needs to follow a standard set of rules. This is where protocols come in. Currently, the main players are ACP, AP2, and Google’s Universal Commerce Protocol (UCP).

These protocols are still evolving, and they solve slightly different parts of the agentic commerce puzzle. Here’s how they compare.

What is ACP (Agentic Commerce Protocol)?

The Agentic Commerce Protocol (ACP) is the standard created by OpenAI and Stripe.

Status: ACP launched in September 2025 through ChatGPT's Instant Checkout, starting with Etsy and Shopify sellers in the US. Take-up stayed low, and OpenAI retired Instant Checkout in March 2026. The protocol itself remains active, but OpenAI’s focus has shifted from a single in-chat buy button towards business-built apps and merchant-controlled checkout experiences inside ChatGPT.
Focus: ACP enables human-present (assisted shopping) flows, allowing a customer to complete a purchase securely within their conversation with an AI agent. It also includes the APIs that let brands expose their inventory and finalise checkouts.
Our strategy: We built Mollie's compatibility with ACP's Delegated Payment API early, while the protocol was still centred on in-chat checkout. We're keeping that infrastructure current as OpenAI's own approach evolves, so you're ready to get paid through whichever ACP-powered surface – in-chat or a merchant app – reaches your customers.

What is AP2 (Agent Payments Protocol 2)?

The Agent Payments Protocol 2 (AP2) is an open-source framework introduced by Google Cloud, Coinbase, and over 60 industry partners to enable secure, verifiable, autonomous payments for AI agents.

AP2 extends the open-source Agent2Agent (A2A) protocol to handle payments, and UCP in turn uses AP2 as its payment layer.

Status: AP2 has released v0.2 and its development has moved under the FIDO Alliance. It’s designed for production use, but widespread live consumer checkouts are still emerging.
Focus: AP2 was designed to support human-absent (delegated shopping) flows from day one. This means the AI can act fully on the customer’s behalf to complete a purchase based on a pre-approved mandate (e.g., “Buy me new coffee beans when I run out”). It also supports human-present (assisted) shopping.
Our strategy: We’re an active partner in the AP2 consortium, helping shape a standard that serves long-term European needs for flexibility and security. We want all payment methods, not only cards, to work in an agentic checkout.

What is UCP (Universal Commerce Protocol)?

The Universal Commerce Protocol (UCP) is an open-source standard co-developed by Google and industry partners to enable agentic commerce across AI platforms, web apps, and business backends.

Status: UCP is live in the US on Google’s AI surfaces, AI Mode in Google Search and Gemini. It’s still an evolving standard, and not every feature in the specification is available on those surfaces yet. Merchants must join a waitlist and be approved by Google before they can go live.
Focus: UCP offers two integration paths. In the native checkout path, customers complete the purchase on Google’s AI surfaces using a Google Pay payment handler, and the business sends order updates after purchase. In the cart path, customers move from Google’s surfaces to the business's website to complete the purchase.
Our strategy: We support all the payment protocols that are made available, and we’re following the industry’s move toward UCP. Our aim is to make sure the standard works for European merchants and consumers, including local payment methods.

ACP, AP2, and UCP: Key differences

Feature	ACP (Agentic Commerce Protocol)	AP2 (Agent Payments Protocol 2)	UCP (Universal Commerce Protocol)
Status	Launched 2025, now powers business apps in ChatGPT	Released as v0.2, now under FIDO Alliance governance; built for production, but live consumer checkouts are still emerging	Live on Google AI surfaces (AI Mode in Search and Gemini), with business approval required
Primary goal	Primarily designed for human-present (assisted) flows	Designed for both human-absent (delegated) and human-present (assisted) flows	Standardises the full purchase process: cart, checkout, and payment
Analogy	A personal shopper helping a customer find and buy items	A personal shopper completes a purchase independently	A shop assistant who handles the whole purchase in one shared format
Key players	OpenAI / Stripe / PayPal (Mollie built compatibility with ACP's Delegated Payment API)	Google / Partners (including Mollie)	Google

For payments to happen securely inside an AI conversation, everyone needs to follow a standard set of rules. This is where protocols come in. Currently, the main players are ACP, AP2, and Google’s Universal Commerce Protocol (UCP).

These protocols are still evolving, and they solve slightly different parts of the agentic commerce puzzle. Here’s how they compare.

What is ACP (Agentic Commerce Protocol)?

The Agentic Commerce Protocol (ACP) is the standard created by OpenAI and Stripe.

Status: ACP launched in September 2025 through ChatGPT's Instant Checkout, starting with Etsy and Shopify sellers in the US. Take-up stayed low, and OpenAI retired Instant Checkout in March 2026. The protocol itself remains active, but OpenAI’s focus has shifted from a single in-chat buy button towards business-built apps and merchant-controlled checkout experiences inside ChatGPT.
Focus: ACP enables human-present (assisted shopping) flows, allowing a customer to complete a purchase securely within their conversation with an AI agent. It also includes the APIs that let brands expose their inventory and finalise checkouts.
Our strategy: We built Mollie's compatibility with ACP's Delegated Payment API early, while the protocol was still centred on in-chat checkout. We're keeping that infrastructure current as OpenAI's own approach evolves, so you're ready to get paid through whichever ACP-powered surface – in-chat or a merchant app – reaches your customers.

What is AP2 (Agent Payments Protocol 2)?

The Agent Payments Protocol 2 (AP2) is an open-source framework introduced by Google Cloud, Coinbase, and over 60 industry partners to enable secure, verifiable, autonomous payments for AI agents.

AP2 extends the open-source Agent2Agent (A2A) protocol to handle payments, and UCP in turn uses AP2 as its payment layer.

Status: AP2 has released v0.2 and its development has moved under the FIDO Alliance. It’s designed for production use, but widespread live consumer checkouts are still emerging.
Focus: AP2 was designed to support human-absent (delegated shopping) flows from day one. This means the AI can act fully on the customer’s behalf to complete a purchase based on a pre-approved mandate (e.g., “Buy me new coffee beans when I run out”). It also supports human-present (assisted) shopping.
Our strategy: We’re an active partner in the AP2 consortium, helping shape a standard that serves long-term European needs for flexibility and security. We want all payment methods, not only cards, to work in an agentic checkout.

What is UCP (Universal Commerce Protocol)?

The Universal Commerce Protocol (UCP) is an open-source standard co-developed by Google and industry partners to enable agentic commerce across AI platforms, web apps, and business backends.

Status: UCP is live in the US on Google’s AI surfaces, AI Mode in Google Search and Gemini. It’s still an evolving standard, and not every feature in the specification is available on those surfaces yet. Merchants must join a waitlist and be approved by Google before they can go live.
Focus: UCP offers two integration paths. In the native checkout path, customers complete the purchase on Google’s AI surfaces using a Google Pay payment handler, and the business sends order updates after purchase. In the cart path, customers move from Google’s surfaces to the business's website to complete the purchase.
Our strategy: We support all the payment protocols that are made available, and we’re following the industry’s move toward UCP. Our aim is to make sure the standard works for European merchants and consumers, including local payment methods.

ACP, AP2, and UCP: Key differences

Feature	ACP (Agentic Commerce Protocol)	AP2 (Agent Payments Protocol 2)	UCP (Universal Commerce Protocol)
Status	Launched 2025, now powers business apps in ChatGPT	Released as v0.2, now under FIDO Alliance governance; built for production, but live consumer checkouts are still emerging	Live on Google AI surfaces (AI Mode in Search and Gemini), with business approval required
Primary goal	Primarily designed for human-present (assisted) flows	Designed for both human-absent (delegated) and human-present (assisted) flows	Standardises the full purchase process: cart, checkout, and payment
Analogy	A personal shopper helping a customer find and buy items	A personal shopper completes a purchase independently	A shop assistant who handles the whole purchase in one shared format
Key players	OpenAI / Stripe / PayPal (Mollie built compatibility with ACP's Delegated Payment API)	Google / Partners (including Mollie)	Google

For payments to happen securely inside an AI conversation, everyone needs to follow a standard set of rules. This is where protocols come in. Currently, the main players are ACP, AP2, and Google’s Universal Commerce Protocol (UCP).

These protocols are still evolving, and they solve slightly different parts of the agentic commerce puzzle. Here’s how they compare.

What is ACP (Agentic Commerce Protocol)?

The Agentic Commerce Protocol (ACP) is the standard created by OpenAI and Stripe.

Status: ACP launched in September 2025 through ChatGPT's Instant Checkout, starting with Etsy and Shopify sellers in the US. Take-up stayed low, and OpenAI retired Instant Checkout in March 2026. The protocol itself remains active, but OpenAI’s focus has shifted from a single in-chat buy button towards business-built apps and merchant-controlled checkout experiences inside ChatGPT.
Focus: ACP enables human-present (assisted shopping) flows, allowing a customer to complete a purchase securely within their conversation with an AI agent. It also includes the APIs that let brands expose their inventory and finalise checkouts.
Our strategy: We built Mollie's compatibility with ACP's Delegated Payment API early, while the protocol was still centred on in-chat checkout. We're keeping that infrastructure current as OpenAI's own approach evolves, so you're ready to get paid through whichever ACP-powered surface – in-chat or a merchant app – reaches your customers.

What is AP2 (Agent Payments Protocol 2)?

The Agent Payments Protocol 2 (AP2) is an open-source framework introduced by Google Cloud, Coinbase, and over 60 industry partners to enable secure, verifiable, autonomous payments for AI agents.

AP2 extends the open-source Agent2Agent (A2A) protocol to handle payments, and UCP in turn uses AP2 as its payment layer.

Status: AP2 has released v0.2 and its development has moved under the FIDO Alliance. It’s designed for production use, but widespread live consumer checkouts are still emerging.
Focus: AP2 was designed to support human-absent (delegated shopping) flows from day one. This means the AI can act fully on the customer’s behalf to complete a purchase based on a pre-approved mandate (e.g., “Buy me new coffee beans when I run out”). It also supports human-present (assisted) shopping.
Our strategy: We’re an active partner in the AP2 consortium, helping shape a standard that serves long-term European needs for flexibility and security. We want all payment methods, not only cards, to work in an agentic checkout.

What is UCP (Universal Commerce Protocol)?

The Universal Commerce Protocol (UCP) is an open-source standard co-developed by Google and industry partners to enable agentic commerce across AI platforms, web apps, and business backends.

Status: UCP is live in the US on Google’s AI surfaces, AI Mode in Google Search and Gemini. It’s still an evolving standard, and not every feature in the specification is available on those surfaces yet. Merchants must join a waitlist and be approved by Google before they can go live.
Focus: UCP offers two integration paths. In the native checkout path, customers complete the purchase on Google’s AI surfaces using a Google Pay payment handler, and the business sends order updates after purchase. In the cart path, customers move from Google’s surfaces to the business's website to complete the purchase.
Our strategy: We support all the payment protocols that are made available, and we’re following the industry’s move toward UCP. Our aim is to make sure the standard works for European merchants and consumers, including local payment methods.

ACP, AP2, and UCP: Key differences

Feature	ACP (Agentic Commerce Protocol)	AP2 (Agent Payments Protocol 2)	UCP (Universal Commerce Protocol)
Status	Launched 2025, now powers business apps in ChatGPT	Released as v0.2, now under FIDO Alliance governance; built for production, but live consumer checkouts are still emerging	Live on Google AI surfaces (AI Mode in Search and Gemini), with business approval required
Primary goal	Primarily designed for human-present (assisted) flows	Designed for both human-absent (delegated) and human-present (assisted) flows	Standardises the full purchase process: cart, checkout, and payment
Analogy	A personal shopper helping a customer find and buy items	A personal shopper completes a purchase independently	A shop assistant who handles the whole purchase in one shared format
Key players	OpenAI / Stripe / PayPal (Mollie built compatibility with ACP's Delegated Payment API)	Google / Partners (including Mollie)	Google

How does agentic commerce differ from traditional ecommerce?

In the traditional ecommerce model, the human shopper does the heavy lifting. They research, compare, and decide. Your website’s job is simply to help a human find what they’re looking for (or suggest things they might like).

With agentic commerce, that dynamic flips.

In assisted flows: The shopper delegates the legwork. The AI researches, filters, and presents the best options, but the human makes the final decision.
In delegated flows: The shopper delegates the decision. The AI acts on its own to buy on the user’s behalf based on pre-set rules.

In agentic flows, software may become the first interface through which customers discover and evaluate your business. That's the critical shift – your business needs to be ready to talk to an AI's API as well as a human browser, since not every customer will want to use agentic commerce.

Aspect	Traditional ecommerce	Agentic commerce (delegated and assisted flow)
The buyer	The human is the decision-maker and executor.	Delegated flow: The AI Agent is the decision-maker and executor. Assisted flow: The AI curates options, the human decides and approves.
The journey	Manual navigation: Search → Browse → Compare → Click → Checkout.	Delegated flow: Goal → Agent research → Agent decision → Purchase. Assisted flow: Goal → AI research & curation → human selection → Purchase.
Visibility	Design-focused: Winning with SEO, advertising, and a beautiful website.	Data-focused: Winning with clean, structured, real-time API data that is machine-readable.
Customer friction	High: manual input of shipping/payment details, multiple steps.	Low where in-chat checkout is available: tokenised payment can reduce checkout to a single conversational step in assisted flows, or remove manual steps in delegated flows. In-chat checkout is still early, and experiences vary.

In the traditional ecommerce model, the human shopper does the heavy lifting. They research, compare, and decide. Your website’s job is simply to help a human find what they’re looking for (or suggest things they might like).

With agentic commerce, that dynamic flips.

In assisted flows: The shopper delegates the legwork. The AI researches, filters, and presents the best options, but the human makes the final decision.
In delegated flows: The shopper delegates the decision. The AI acts on its own to buy on the user’s behalf based on pre-set rules.

In agentic flows, software may become the first interface through which customers discover and evaluate your business. That's the critical shift – your business needs to be ready to talk to an AI's API as well as a human browser, since not every customer will want to use agentic commerce.

Aspect	Traditional ecommerce	Agentic commerce (delegated and assisted flow)
The buyer	The human is the decision-maker and executor.	Delegated flow: The AI Agent is the decision-maker and executor. Assisted flow: The AI curates options, the human decides and approves.
The journey	Manual navigation: Search → Browse → Compare → Click → Checkout.	Delegated flow: Goal → Agent research → Agent decision → Purchase. Assisted flow: Goal → AI research & curation → human selection → Purchase.
Visibility	Design-focused: Winning with SEO, advertising, and a beautiful website.	Data-focused: Winning with clean, structured, real-time API data that is machine-readable.
Customer friction	High: manual input of shipping/payment details, multiple steps.	Low where in-chat checkout is available: tokenised payment can reduce checkout to a single conversational step in assisted flows, or remove manual steps in delegated flows. In-chat checkout is still early, and experiences vary.

In the traditional ecommerce model, the human shopper does the heavy lifting. They research, compare, and decide. Your website’s job is simply to help a human find what they’re looking for (or suggest things they might like).

With agentic commerce, that dynamic flips.

In assisted flows: The shopper delegates the legwork. The AI researches, filters, and presents the best options, but the human makes the final decision.
In delegated flows: The shopper delegates the decision. The AI acts on its own to buy on the user’s behalf based on pre-set rules.

In agentic flows, software may become the first interface through which customers discover and evaluate your business. That's the critical shift – your business needs to be ready to talk to an AI's API as well as a human browser, since not every customer will want to use agentic commerce.

Aspect	Traditional ecommerce	Agentic commerce (delegated and assisted flow)
The buyer	The human is the decision-maker and executor.	Delegated flow: The AI Agent is the decision-maker and executor. Assisted flow: The AI curates options, the human decides and approves.
The journey	Manual navigation: Search → Browse → Compare → Click → Checkout.	Delegated flow: Goal → Agent research → Agent decision → Purchase. Assisted flow: Goal → AI research & curation → human selection → Purchase.
Visibility	Design-focused: Winning with SEO, advertising, and a beautiful website.	Data-focused: Winning with clean, structured, real-time API data that is machine-readable.
Customer friction	High: manual input of shipping/payment details, multiple steps.	Low where in-chat checkout is available: tokenised payment can reduce checkout to a single conversational step in assisted flows, or remove manual steps in delegated flows. In-chat checkout is still early, and experiences vary.

In the traditional ecommerce model, the human shopper does the heavy lifting. They research, compare, and decide. Your website’s job is simply to help a human find what they’re looking for (or suggest things they might like).

With agentic commerce, that dynamic flips.

In assisted flows: The shopper delegates the legwork. The AI researches, filters, and presents the best options, but the human makes the final decision.
In delegated flows: The shopper delegates the decision. The AI acts on its own to buy on the user’s behalf based on pre-set rules.

In agentic flows, software may become the first interface through which customers discover and evaluate your business. That's the critical shift – your business needs to be ready to talk to an AI's API as well as a human browser, since not every customer will want to use agentic commerce.

Aspect	Traditional ecommerce	Agentic commerce (delegated and assisted flow)
The buyer	The human is the decision-maker and executor.	Delegated flow: The AI Agent is the decision-maker and executor. Assisted flow: The AI curates options, the human decides and approves.
The journey	Manual navigation: Search → Browse → Compare → Click → Checkout.	Delegated flow: Goal → Agent research → Agent decision → Purchase. Assisted flow: Goal → AI research & curation → human selection → Purchase.
Visibility	Design-focused: Winning with SEO, advertising, and a beautiful website.	Data-focused: Winning with clean, structured, real-time API data that is machine-readable.
Customer friction	High: manual input of shipping/payment details, multiple steps.	Low where in-chat checkout is available: tokenised payment can reduce checkout to a single conversational step in assisted flows, or remove manual steps in delegated flows. In-chat checkout is still early, and experiences vary.

What are agentic commerce use cases and opportunities?

Agentic commerce isn’t just a new channel; it’s a more efficient way to connect supply and demand. Here’s just a taste of what it offers:

New reach: AI agents operate on platforms used by millions of users (like ChatGPT and Google Assistant), opening up entirely new sales opportunities for your business.
Smarter product discovery: AI agents can act as highly capable personal shoppers, matching products to specific, complex customer requests (e.g., “Find me a sustainable, size 10, waterproof rain jacket for under €150 and order it.”).
Potential for faster conversion: In theory, the entire customer journey, from search to product discovery to purchase, can happen within a single conversation. This could reduce checkout friction and drop-off compared to traditional, multi-step ecommerce paths.
Simpler checkout (where available): With tokenised payments, the customer never has to enter payment or shipping details, making the purchase lightning-fast.

Agentic commerce isn’t just a new channel; it’s a more efficient way to connect supply and demand. Here’s just a taste of what it offers:

New reach: AI agents operate on platforms used by millions of users (like ChatGPT and Google Assistant), opening up entirely new sales opportunities for your business.
Smarter product discovery: AI agents can act as highly capable personal shoppers, matching products to specific, complex customer requests (e.g., “Find me a sustainable, size 10, waterproof rain jacket for under €150 and order it.”).
Potential for faster conversion: In theory, the entire customer journey, from search to product discovery to purchase, can happen within a single conversation. This could reduce checkout friction and drop-off compared to traditional, multi-step ecommerce paths.
Simpler checkout (where available): With tokenised payments, the customer never has to enter payment or shipping details, making the purchase lightning-fast.

Agentic commerce isn’t just a new channel; it’s a more efficient way to connect supply and demand. Here’s just a taste of what it offers:

New reach: AI agents operate on platforms used by millions of users (like ChatGPT and Google Assistant), opening up entirely new sales opportunities for your business.
Smarter product discovery: AI agents can act as highly capable personal shoppers, matching products to specific, complex customer requests (e.g., “Find me a sustainable, size 10, waterproof rain jacket for under €150 and order it.”).
Potential for faster conversion: In theory, the entire customer journey, from search to product discovery to purchase, can happen within a single conversation. This could reduce checkout friction and drop-off compared to traditional, multi-step ecommerce paths.
Simpler checkout (where available): With tokenised payments, the customer never has to enter payment or shipping details, making the purchase lightning-fast.

Agentic commerce isn’t just a new channel; it’s a more efficient way to connect supply and demand. Here’s just a taste of what it offers:

New reach: AI agents operate on platforms used by millions of users (like ChatGPT and Google Assistant), opening up entirely new sales opportunities for your business.
Smarter product discovery: AI agents can act as highly capable personal shoppers, matching products to specific, complex customer requests (e.g., “Find me a sustainable, size 10, waterproof rain jacket for under €150 and order it.”).
Potential for faster conversion: In theory, the entire customer journey, from search to product discovery to purchase, can happen within a single conversation. This could reduce checkout friction and drop-off compared to traditional, multi-step ecommerce paths.
Simpler checkout (where available): With tokenised payments, the customer never has to enter payment or shipping details, making the purchase lightning-fast.

How do I prepare for agentic commerce’s core challenges?

Agentic commerce introduces two strategic challenges your business must prepare for. Building trust means being transparent about the required shift.

The brand and retention challenge

In traditional ecommerce, you own the full customer journey. In agentic commerce, your customer acquisition is outsourced to an AI. That first sale, executed by a bot, is impersonal by default.

Your most critical challenge will be converting that first-time AI-acquired customer into a loyal direct customer with a real post-sale relationship. Your customer service, fulfilment, and post-purchase experience will be more vital than ever for long-term value.

The technical preparation challenge

To be ‘found’ and have your products purchased by an AI agent, your business information must be machine-readable. AI Agents also prefer product information in text and structured data. If your sizes, materials, or stock levels appear only in images, an agent may never surface your product. Optimising for agents, not just search engines, is called generative engine optimisation (GEO).

This means all crucial business information must be accessible via APIs, not just as text on your website. This includes product descriptions, stock levels, and sizing. But it also includes your trust signals: your return windows, shipping guarantees, and refund policies. An FAQ page alone is not enough – agents need accurate, structured, machine-readable information about returns, delivery, stock, and policies before they can confidently recommend or purchase.

You do not need to rebuild your store overnight. But you can start making it easier for AI agents to understand, trust, and transact with your business.

Start with these things:

Make product data machine-readable. Put sizes, materials, stock levels, and key product attributes in structured text and data, not only images.
Make policies easy to verify. Ensure returns, delivery, refunds, and shipping information are accurate, current, and available in structured formats.
Keep your ecommerce setup healthy. Update your platform, plugins, and integrations to reduce technical issues that could affect discoverability.
Monitor your AI visibility. Search for your brand and products in AI assistants, compare the results with competitors, and identify gaps in how your business is described.

Your store will serve two kinds of visitors: people and agents. Agents can handle more detail, so you can add structured information for them while keeping pages clear for human customers.

Agentic commerce introduces two strategic challenges your business must prepare for. Building trust means being transparent about the required shift.

The brand and retention challenge

In traditional ecommerce, you own the full customer journey. In agentic commerce, your customer acquisition is outsourced to an AI. That first sale, executed by a bot, is impersonal by default.

Your most critical challenge will be converting that first-time AI-acquired customer into a loyal direct customer with a real post-sale relationship. Your customer service, fulfilment, and post-purchase experience will be more vital than ever for long-term value.

The technical preparation challenge

To be ‘found’ and have your products purchased by an AI agent, your business information must be machine-readable. AI Agents also prefer product information in text and structured data. If your sizes, materials, or stock levels appear only in images, an agent may never surface your product. Optimising for agents, not just search engines, is called generative engine optimisation (GEO).

This means all crucial business information must be accessible via APIs, not just as text on your website. This includes product descriptions, stock levels, and sizing. But it also includes your trust signals: your return windows, shipping guarantees, and refund policies. An FAQ page alone is not enough – agents need accurate, structured, machine-readable information about returns, delivery, stock, and policies before they can confidently recommend or purchase.

You do not need to rebuild your store overnight. But you can start making it easier for AI agents to understand, trust, and transact with your business.

Start with these things:

Make product data machine-readable. Put sizes, materials, stock levels, and key product attributes in structured text and data, not only images.
Make policies easy to verify. Ensure returns, delivery, refunds, and shipping information are accurate, current, and available in structured formats.
Keep your ecommerce setup healthy. Update your platform, plugins, and integrations to reduce technical issues that could affect discoverability.
Monitor your AI visibility. Search for your brand and products in AI assistants, compare the results with competitors, and identify gaps in how your business is described.

Your store will serve two kinds of visitors: people and agents. Agents can handle more detail, so you can add structured information for them while keeping pages clear for human customers.

Agentic commerce introduces two strategic challenges your business must prepare for. Building trust means being transparent about the required shift.

The brand and retention challenge

In traditional ecommerce, you own the full customer journey. In agentic commerce, your customer acquisition is outsourced to an AI. That first sale, executed by a bot, is impersonal by default.

Your most critical challenge will be converting that first-time AI-acquired customer into a loyal direct customer with a real post-sale relationship. Your customer service, fulfilment, and post-purchase experience will be more vital than ever for long-term value.

The technical preparation challenge

To be ‘found’ and have your products purchased by an AI agent, your business information must be machine-readable. AI Agents also prefer product information in text and structured data. If your sizes, materials, or stock levels appear only in images, an agent may never surface your product. Optimising for agents, not just search engines, is called generative engine optimisation (GEO).

This means all crucial business information must be accessible via APIs, not just as text on your website. This includes product descriptions, stock levels, and sizing. But it also includes your trust signals: your return windows, shipping guarantees, and refund policies. An FAQ page alone is not enough – agents need accurate, structured, machine-readable information about returns, delivery, stock, and policies before they can confidently recommend or purchase.

You do not need to rebuild your store overnight. But you can start making it easier for AI agents to understand, trust, and transact with your business.

Start with these things:

Make product data machine-readable. Put sizes, materials, stock levels, and key product attributes in structured text and data, not only images.
Make policies easy to verify. Ensure returns, delivery, refunds, and shipping information are accurate, current, and available in structured formats.
Keep your ecommerce setup healthy. Update your platform, plugins, and integrations to reduce technical issues that could affect discoverability.
Monitor your AI visibility. Search for your brand and products in AI assistants, compare the results with competitors, and identify gaps in how your business is described.

Your store will serve two kinds of visitors: people and agents. Agents can handle more detail, so you can add structured information for them while keeping pages clear for human customers.

Agentic commerce introduces two strategic challenges your business must prepare for. Building trust means being transparent about the required shift.

The brand and retention challenge

In traditional ecommerce, you own the full customer journey. In agentic commerce, your customer acquisition is outsourced to an AI. That first sale, executed by a bot, is impersonal by default.

Your most critical challenge will be converting that first-time AI-acquired customer into a loyal direct customer with a real post-sale relationship. Your customer service, fulfilment, and post-purchase experience will be more vital than ever for long-term value.

The technical preparation challenge

To be ‘found’ and have your products purchased by an AI agent, your business information must be machine-readable. AI Agents also prefer product information in text and structured data. If your sizes, materials, or stock levels appear only in images, an agent may never surface your product. Optimising for agents, not just search engines, is called generative engine optimisation (GEO).

This means all crucial business information must be accessible via APIs, not just as text on your website. This includes product descriptions, stock levels, and sizing. But it also includes your trust signals: your return windows, shipping guarantees, and refund policies. An FAQ page alone is not enough – agents need accurate, structured, machine-readable information about returns, delivery, stock, and policies before they can confidently recommend or purchase.

You do not need to rebuild your store overnight. But you can start making it easier for AI agents to understand, trust, and transact with your business.

Start with these things:

Make product data machine-readable. Put sizes, materials, stock levels, and key product attributes in structured text and data, not only images.
Make policies easy to verify. Ensure returns, delivery, refunds, and shipping information are accurate, current, and available in structured formats.
Keep your ecommerce setup healthy. Update your platform, plugins, and integrations to reduce technical issues that could affect discoverability.
Monitor your AI visibility. Search for your brand and products in AI assistants, compare the results with competitors, and identify gaps in how your business is described.

Your store will serve two kinds of visitors: people and agents. Agents can handle more detail, so you can add structured information for them while keeping pages clear for human customers.

How does agentic commerce change security and fraud?

Agentic commerce introduces new security considerations, particularly as AI systems begin to initiate and complete transactions.

New AI-driven fraud threats

Fraudsters will use AI to create new threats to businesses.

New fraud patterns: AI can help fraudsters test and adapt attack patterns much faster than before, which puts pressure on static, rules-only fraud controls.
Believable fake behaviour: Malicious bots can execute highly believable fake customer behaviours at scale, making it difficult to distinguish real customers from automated attackers.

The real-time defence imperative

You need to delegate your real-time defence to your payment partner. Mollie is addressing this by:

Tokenisation: Using the Shared Payment Token system ensures raw payment data is never exposed.
Adaptive models: We combine two layers of defence. First, our own AI models – trained on millions of European transactions – detect fraud patterns in real time across all our transactions, agentic or not. This is how we verify the identity behind a standard transaction, not just the card number. Second, for agentic transactions specifically, we integrate risk scores and behavioural indicators provided directly by the AI platforms. Verifying the identity of the agent itself, not just the customer behind it, is a harder problem – one the whole industry, including the standards bodies behind AP2 and the Five Eyes agentic AI guidance, is still working through.

In May 2026, six national cybersecurity agencies – including the US's CISA and NSA, and counterparts in Australia, Canada, New Zealand and the UK – published joint guidance on securing agentic AI systems. Their core recommendation is least-privilege access: agents should get the smallest set of permissions a task needs, with fail-safe defaults that make an agent stop and ask rather than guess. That principle is already shaping how the industry approaches agentic payments.

Two concepts are emerging to protect merchants from bots. Verifiable intent is a new approach that checks whether a purchase reflects a real intention to buy, rather than an AI hallucination. Know your agent is the idea that agents, like customers, should be identifiable and accountable. Both are worth following as the standards develop.

How do agentic commerce payments work?

The agentic commerce payment process relies on tokenisation. This ensures the transaction is secure and that the AI assistant never touches sensitive data.

A secure, token-based payment flow

Agentic commerce payments are highly secure and rely on tokenisation:

No sharing of raw data: Raw payment credentials, such as card numbers, are not shared with the AI assistant. The agent passes a scoped payment token or payment reference to complete the transaction.
Token generation: The payment provider (like Mollie) generates a secure, single-use, scoped Shared Payment Token for the transaction.
Payment processing: This token is what the AI agent passes to the business's system to complete the transaction.
Accountability: Agents need to tell merchants, card schemes, and payment methods which customer they act for, and keep a record of each decision. This trail shows who made a mistake and who is liable, which protects both merchants and customers.

This system aims to keep European compliance requirements, including Strong Customer Authentication (SCA), in place while maintaining transaction security and customer trust. Many agentic flows don't yet support the second-factor confirmation that SCA requires mid-purchase. Mollie is working on making SCA work in agentic checkout.

Why local payment methods are the key to unlocking Europe

The current protocol standards are initially US-centric and support only global, card-based methods (Apple Pay, Google Pay, Cards). This model is incomplete for European success. A Dutch customer expects iDEAL | Wero; a Belgian customer expects Bancontact. Forcing them to use a card in a chat is not frictionless – it is a critical point of failure.

Our core mission is to represent the needs of European businesses and help ensure these new global protocols support the local payment methods your customers actually use and trust.

Businesses shouldn’t have to choose between offering local payment methods and being visible to AI agents. You can have both.

How can a local payment method (like iDEAL | Wero) even work in a chat?

This is one of the most complex challenges in agentic commerce. Bank redirects and app-to-app payments do not naturally fit a conversational interface, so the user experience needs careful design.

Our product and engineering teams are actively designing and testing solutions to streamline this flow, ensuring the process remains secure, compliant, and – most importantly – effortless for the end user.

One option is to store payment details in an account the customer already trusts, so the agent can complete a purchase with the customer’s permission. The industry is still working out how this kind of delegated authority should work.

What is Mollie’s strategy for agentic commerce?

The agentic commerce landscape can appear complex, but your payments don’t have to be.

A practical path through a changing landscape

You should not need to become an expert in competing protocols. Mollie’s role is to help you stay ready as the standards evolve.

Stay ready across ACP's evolving model: we built Delegated Payment API support for ACP before OpenAI narrowed its focus to merchant apps inside ChatGPT. We're keeping that infrastructure current as the protocol's role shifts, so you're not left reintegrating every time it changes.
Shape the future with AP2: We’re an active partner in the AP2 consortium, collaborating with Google to help shape a future-proof standard that serves European needs.

Businesses can get ready for agentic commerce with four steps:

Check that your payment provider supports the protocols you sell through. Mollie is compatible with ACP, and we’ll add support as the standards develop.
Make your product data available through APIs, not only on web pages. Stock, sizes, and return policies should be readable by agents.
Keep your ecommerce platform and plugins up to date. Updates help your products stay discoverable as the standards change.
Plan the post-purchase experience. It’s often the first time a customer meets your brand directly, so make it count.

Get ready for what comes next

Agentic commerce is still taking shape, but the direction is clear: AI is becoming a new interface for discovery, comparison, and purchase.

Mollie’s role is to handle the payments complexity, so you can focus on running your business. We are building for the protocols emerging today and working to ensure European businesses can get paid securely, compliantly, and with the local payment methods their customers expect.

Want to join us in this new frontier? Explore online payments with Mollie.

banner payments report 2026

banner payments report 2026

The Payments Report 2026

Get the operational playbook to automate complexity and turn transaction data into revenue.

Download the report

Download the report

More updates

What is Pay by Bank and how does it work?

Explore how Pay by Bank and Open Banking payments work and the benefits they offer to both businesses and customers. Offer Pay by Bank today.

Bringing your Pickware payments together with Mollie: one partner for every channel

You’ve unified your business with Pickware. See how Mollie unifies your online and in-store payments too.

Wero dispute handling process: what businesses need to know

Wero introduces disputes and chargebacks that guaranteed payment methods never had. Here's how it works, what it costs, and how to protect your margin.

What is Click to Pay? A guide for ecommerce companies

Learn how Click to Pay uses tokenisation to simplify online card payments for customers and what it could mean for your checkout.

Agentic commerce: your questions answered

Our payment infrastructure is ready for the future of AI sales. Now, read our expert guide to prepare for the agentic commerce era.

What is Pay by Bank and how does it work?

Explore how Pay by Bank and Open Banking payments work and the benefits they offer to both businesses and customers. Offer Pay by Bank today.

Bringing your Pickware payments together with Mollie: one partner for every channel

You’ve unified your business with Pickware. See how Mollie unifies your online and in-store payments too.

Wero dispute handling process: what businesses need to know

Wero introduces disputes and chargebacks that guaranteed payment methods never had. Here's how it works, what it costs, and how to protect your margin.

Table of contents

Table of contents

Partner featured

Table of contents

What this means for your business

What does the agentic ecosystem look like?

What are the competing agentic commerce protocols?

How does agentic commerce differ from traditional ecommerce?

What are agentic commerce use cases and opportunities?

How do I prepare for agentic commerce’s core challenges?

How does agentic commerce change security and fraud?

How do agentic commerce payments work?

Why local payment methods are the key to unlocking Europe

What is Mollie’s strategy for agentic commerce?

Get ready for what comes next

Contact sales

Mollie GrowthAgentic commerce: your questions answered

Mollie GrowthAgentic commerce: your questions answered

Mollie GrowthAgentic commerce: your questions answered