What is agentic check-out?

Agentic check-out lets shoppers make purchases directly within AI agent conversations. This process can happen in an AI agent chat interface like Gemini, via third-party AI agents, or with an AI agent on a retailer’s website or ecommerce storefront.

Agentic check-out looks something like this: a shopper chats with an AI agent and describes a specific need or delegates a purchasing task. The AI agent surfaces product listings, and the shopper completes the purchase either within the chat interface or through a third party.

The AI agent enables the purchase by recommending products, then helping the shopper add items to a cart and complete the purchase.

How does agentic check-out work?

Agentic check-out simplifies the shopping experience for consumers, but a lot happens behind the scenes to make it look effortless. Here’s how it works:

How agentic check-out works for shoppers

1. The shopper searches for products by having a conversation with an AI agent. For example, a business traveller might ask ChatGPT for a suitcase that fits in the overhead bins and has lots of pockets for organisation. A customer using a brand’s AI agent might prompt it to reorder their vitamins once they run out.
2. If the search is happening with an external AI platform like ChatGPT or Gemini, the platform queries merchants using open commerce protocols such as Agentic Commerce Protocol (ACP) or Universal Commerce Protocol (UCP) that standardise how AI-assisted purchases are discovered, checked out, and paid for.
3. The AI agent identifies relevant products, checks pricing and availability, and finds current promotions. Most AI interactions are personalised based on past conversations or customer data, so listings may also bring up this context.
4. If the shopper finds what they’re looking for, they can ask the AI agent to place the item in a cart or click through to make a purchase.

How agentic check-out works for merchants

1. You can adopt the ACP and UCP protocols to enable agentic check-out through external AI agents. You can also use a branded AI customer agent on your website to support agentic check-out.
2. When the shopper is ready to buy, the AI agent facilitates payment through secure, tokenised transactions. You receive the order and fulfil it as you normally would, with the customer data and attribution that comes with it.
3. Brands using Shopify and Klaviyo will see agentic shopping sources show up in their attribution data if their US-based stores are eligible. You can use sources to create flows such as abandoned cart reminders, customer segments by channel and product category, and campaigns targeted to lifecycle stage.

Where is agentic check-out available today?

Agentic check-out is available across multiple AI platforms and payment vendors, such as:

• Google: The UCP powers agentic commerce in Google AI Search Mode and Gemini.
• ChatGPT: Powered by the ACP and Stripe, shoppers can discover products and build their carts with the AI agent, then complete transactions either on the merchant’s website or inside the merchant’s dedicated app.
• Microsoft Copilot: The ACP powers AI agent-driven purchase workflows in Microsoft Copilot.
• Perplexity: In May 2025, Perplexity partnered with PayPal and Venmo to launch Instant Buy, an agentic check-out option for US users.
• OpenClaw: OpenClaw is an open-source agentic AI assistant people can configure to handle all kinds of tasks, from managing their inbox to ordering a new outfit. While there are still some security concerns for these tools, they may gain popularity in the next few years if people begin to trust AI more.

How is agentic check-out changing the customer experience?

Agentic check-out is changing how consumers and brands interact throughout the entire customer journey, not just at the point of sale.

If you’re just starting to get a handle on agentic check-out and want to know where to focus your energy, consider the following:

1. Brands need to build more credibility

Klaviyo’s 2026 AI Consumer Trends Report found that 60% of global consumers use AI at least weekly, but only 13% of consumers completely trust it. Meanwhile, a 2025 Bain survey on agentic commerce found that only 24% of consumers trust AI to make purchases for them, mainly due to security and privacy concerns.

As more purchases happen outside of brand-owned channels, you need to communicate early and often to maintain customer trust. Do this by:

• Updating and communicating clear data privacy policies that speak to agentic check-out
• Auditing your product descriptions and brand voice for detail and consistency, so you continue to show up across agentic channels in a cohesive way
• Building a human option into every service touchpoint you control, so shoppers can escalate complex questions to a human agent

2. AI shopping prompts are intent and context signals

According to Klaviyo’s 2026 AI Consumer Trends Report, 30% of AI searches contain 8+ keywords. 78% of people say they include emotional or personal context at least some of the time when using AI.

These long-tail prompts are filled with intent and contextual data that reveal much more about your audience than the keywords they traditionally Googled. While you won’t be able to see exactly which prompts generated traffic or purchases, some AI search monitoring tools cluster AI prompts into topics to give you a better sense of how you’re showing up in conversations with AI agents.

For instance, with AI search monitoring and UTM tracking, you can create a custom flow for shoppers who discovered your brand through an AI agent and abandoned a cart. You know these flows will need to contain more context than ones triggered by abandoned carts that came through other channels.

3. Full-funnel attribution is more complicated

82% of consumers rely on more than one channel to shop, according to Klaviyo’s 2026 future of consumer marketing research. Omnichannel shopping complicates attribution enough, but zero-click agentic commerce stands to make it even more complex.

As more transactions are completed off-site, last-click attribution loses visibility into a growing percentage of transactions. Measuring the performance of AI searches means pulling together order data, multi-touch attribution, and webhook-delivered order events from protocol-mediated check-outs.

How can brands optimise the agentic check-out experience?

To make the most of agentic check-out, take these steps:

• Make your products discoverable by AI. Audit your structured product metadata to ensure that descriptions are detailed and current, and that they contain enough long-tail information for AI agents to adapt them for personalised shopping recommendations. By default, AI bot crawlers adhere to your site’s robots.txt file. If you want your product information to be indexed, you need to make sure LLMs can access it.
• Review your current vendors’ AI readiness. Contact your tech providers to assess whether they can support agentic check-out. Use payment platforms that are compatible with agentic check-out, like American Express, Google Pay, Visa, and others.
• Offer customers the same perks as a traditional check-out experience. Regardless of where agentic transactions take place, customers should have the same access to discount codes and loyalty programme benefits. To offer these benefits through agentic check-out, you may need to set up additional infrastructure, such as UCP identity linking.
• Nurture customers after the sale. Invest even more in post-purchase nurture channels like email and text messaging to educate first-time customers and build loyalty. These channels may need to collect more descriptive details for reviews or even introduce your brand if most of the customer’s interactions have taken place off-site.

Is your brand ready for agentic check-out?

Redefine the customer experience with Customer Agent, a personalised shopping assistant that can recommend products, answer questions instantly, and nudge shoppers toward purchase.

Like Composer, Klaviyo’s AI marketing agent, Customer Agent is built directly into Klaviyo B2C CRM, which means it has full access to the customer data, product catalog, and brand guidelines it needs to curate and personalize the customer experience for each individual person. Every outcome feeds back into the system, so each decision Klaviyo’s AI agents make is smarter than the last.