In-Chat Checkout Integration Guide 2026: How You Connect and Get Paid on Perplexity Instant Buy, ChatGPT Apps, and Copilot Checkout

This is the plumbing comparison, not the traffic one

There are already plenty of pieces comparing AI shopping channels on who has the most users and the best conversion. We have a separate article for that fight. This one stays out of the volume conversation entirely and answers two much more practical questions: how do I, as a merchant, wire my store into these systems, and how does the money actually land in my account.

Those questions sound dry, but they are what determines whether you can ship this at all. A channel can have enormous reach and still be a poor fit if connecting it means rebuilding your checkout, or if you are not the merchant-of-record, or if refunds and tax all fall back on you with no support. So below I line up Perplexity Instant Buy, ChatGPT Apps, and Microsoft Copilot Checkout across the dimensions that actually decide integration: how you plug in, the payment rail, who stays merchant-of-record, what catalog you supply, and which platforms are covered.

I will keep the wiring explanation plain, and close with a decision table organized by your stack rather than by hype.

Perplexity Instant Buy: PayPal as the payment layer, free in the US

Perplexity Instant Buy is its in-chat checkout for the US market, and it is free for shoppers to use. The payment layer is built on PayPal, so the transaction settles through the PayPal rail. The merchant-facing logic is straightforward: a shopper states what they want in the conversation, Perplexity generates product cards tailored to that stated intent, and if something fits, the shopper can buy it without leaving the chat.

The integration bar is comparatively low. You are mainly supplying your product information so Perplexity can surface the right cards in the right conversational moments. Because PayPal handles the payment leg, sellers already collecting through PayPal get a rail that is effectively pre-built, with no separate payment infrastructure to stand up.

The thing to keep in mind is that this is a US-market capability today. If your core audience is in the US and PayPal is already part of how you collect, the integration cost is essentially getting your product data connected cleanly. If your settlement runs primarily through another processor, weigh how adding the PayPal layer affects your reconciliation and accounting flow before committing.

ChatGPT Apps: built on ACP, and you stay merchant-of-record

ChatGPT made a clear shift in 2026. The earlier approach leaned toward a more native checkout, and over the year it moved toward the merchant-controlled ChatGPT Apps model. Underneath sits the Agentic Commerce Protocol (ACP), co-developed by Stripe and OpenAI.

For a merchant, the integration path is ACP plus Stripe. One design choice matters enormously here: you remain the merchant-of-record, not OpenAI. In legal and accounting terms, your store is still the one selling the goods. OpenAI provides the conversational surface and the protocol layer, and Stripe provides the payment rail, but the sale is recorded against you.

That arrangement gives you more control over transaction data, the customer relationship, refunds, and tax handling, because you are still the merchant-of-record. The trade-off is real integration work on your side to wire up ACP and Stripe, which is heavier than simply feeding in a product card. For DTC brands already running on Stripe that care about owning their transaction data and customer relationships, that control is the strongest argument for this path.

Microsoft Copilot Checkout: large reach across Shopify merchants

Copilot Checkout stands out for the size of the merchant base it can reach. A commonly cited figure is that it can reach 500,000 or more Shopify merchants. The wording matters: this is a reach figure, not a count of merchants who have integrated or are actively transacting through it, so do not read the number as adoption. Beyond Shopify, it has also extended toward WooCommerce.

The upside is that if you are already on Shopify, the path to being reachable is relatively smooth, with no need to build a connector from scratch. For WooCommerce sellers, the extending support adds another doorway into conversational checkout.

That said, being reachable and being integrated-and-live are two different things. The specifics of the payment rail, the merchant-of-record arrangement, and catalog sync should be confirmed against the actual configuration the platform gives you at integration time. My read: Copilot Checkout has a natural coverage advantage for Shopify-stack sellers, but confirm the settlement and accounting details before you enable it, rather than assuming the 500,000 figure means you are already in the game.

The three side by side: integration, payment, and who carries the risk

Here are the three systems flattened across the dimensions that decide the build:

DimensionPerplexity Instant BuyChatGPT Apps (ACP)Microsoft Copilot Checkout
Integration pathConnect product data, free to startWire up ACP + StripeEnable via the reachable platform doorway
Payment railPayPal as the payment layerStripePer the platform configuration
Merchant-of-recordPer platform configurationStays the merchantConfirm at integration
Catalog / feed requirementFeed product data for intent cardsProduct catalog via ACPLeans on existing platform product data
Platform supportUS market firstSellers running on StripeShopify (500K+ reachable), extending to WooCommerce
Data / controlPerplexity drives card presentationHigh merchant control (still MoR)Platform-dependent, confirm data ownership

Lined up this way, the three point in noticeably different directions. Perplexity leads with a low bar and a free start, but presentation and payment lean more on the platform. ChatGPT Apps hands control back to the merchant, you remain merchant-of-record, and the cost is a heavier integration. Copilot leads on coverage, especially for the Shopify stack, but the settlement specifics need to be confirmed line by line.

One thing worth flagging: who the merchant-of-record is directly decides who carries refunds, disputes, and tax. ChatGPT Apps is explicit that it stays the merchant, which is a plus for brands that care about accounting and customer relationships. For the other two, read the arrangement carefully in the integration terms rather than going on assumption.

Which to prioritize, based on your stack

You do not need all three. Ranking them by your current stack is the least painful way in.

If you are a Shopify merchant: Copilot Checkout’s reach is the most natural fit, the coverage is already there, so confirm your integration and settlement configuration and get that relatively smooth path running first. If you are also on Stripe, put ChatGPT Apps next on the list, because you stay merchant-of-record and the control is strong.

If you are a WooCommerce merchant: watch the progress of Copilot Checkout’s extension toward WooCommerce, since that is a ready-made doorway into conversational checkout. In parallel, assess the integration cost of the ACP plus Stripe path against your existing checkout logic.

If you run a custom or self-hosted stack: ChatGPT Apps via ACP plus Stripe is the most direct fit, since you control the merchant-of-record role and your transaction data, which suits DTC brands that care about data ownership. Perplexity Instant Buy fits when your core audience is in the US and you already collect through PayPal, where the integration cost is mostly getting your product data connected.

My overall advice: pick the single path that best matches your existing payment and platform stack, get settlement and accounting fully validated, and only then decide whether to extend to a second. With in-chat checkout, reliably getting paid and knowing who carries refunds and tax matters far more than spreading across three channels on day one.

Related Articles