NewsCommerce
Perform.AI launches a free checkout readiness checker for AI shopping
Perform.AI’s free tool checks UCP and ACP storefront readiness. See what the results tell merchants about catalog, checkout and order updates.
- Discovery
- Can an agent find the item?
- Checkout
- Can it start the purchase?
- After the sale
- Can it track order updates?
- Connectivity
- Can the systems exchange data?
Perform.AI describes 147 checks across the shopping journey. A score alone does not establish eligibility to sell through a platform.[1][2]
What this means for founders
- Perform.AI announced a free checker on November 18 that evaluates public storefront signals against UCP, ACP and related standards.
- Prioritize missing information that would block a purchase, such as inconsistent stock or an unclear delivery promise.
- Before building a new integration, verify that a shopping platform you can actually reach offers a route to merchants like you.
What launched
Perform.AI announced its UCP and ACP Readiness Checker on November 18, 2026. The tool takes a store domain and reports how much of the shopping journey a software agent can understand from that storefront. Perform.AI says it runs 147 checks covering discovery, checkout, post-purchase updates and technical connectivity. Its launch release says the score is free and that the full report asks for contact details.
The release describes a rules-based scan, with a score out of 100 and separate findings for issues that could block a sale. Its product documentation says that checks it cannot perform from outside the merchant's systems are marked as untested. A merchant can use the output to assemble a list of questions for its ecommerce platform or engineering team.
Two protocols, several paths to a purchase
The Universal Commerce Protocol (UCP) and the Agentic Commerce Protocol (ACP) describe ways for merchants and shopping systems to exchange information and conduct transactions. Google has described optional UCP features for retrieving current product variants, prices and inventory, keeping a cart and linking a customer's account. OpenAI introduced ACP with Stripe in its September 2025 checkout announcement. Each platform determines its own eligibility and supports particular capabilities.
The launch comes days after OpenAI announced an initial US rollout of ChatGPT Wallet for eligible adult Pro subscribers with eligible cards. The Wallet announcement also describes working with existing merchant checkout flows. A merchant evaluating the checker should first identify the shopping experience they are trying to support. One experience might rely on a product feed and a handoff to the store. Another might use parts of a commerce protocol. Those are different projects with different owners and launch requirements.
What a storefront scan can tell you
A useful finding is one you can reproduce against a real transaction. Say your product page says a jacket is in stock, yet the selected size is unavailable at checkout. That inconsistency matters to a shopper using a browser as much as it matters to an assistant comparing options. Fixing it can be useful even when nobody has measured incremental sales from assistants.
The score itself cannot tell you how often a platform will show the store, whether its buyers will convert or whether the orders will be profitable. An external scan also has limited visibility into payment authorization, fulfillment exceptions and the way a particular merchant resolves returns. Use its findings alongside your own catalog records, checkout logs and support tickets.
A practical first test for a small commerce team
Pick one product with variants, one product that frequently goes out of stock and one item with special shipping terms. For each, record which price, quantity, delivery estimate and return policy a buyer sees on the product page and receives at checkout. Give each mismatch a named owner before deciding whether a new integration deserves the next sprint.
- Catalog owners should confirm that the chosen product variant maps to the correct price and availability.
- Checkout owners should exercise an authorized test transaction and record how declines, out-of-stock items and substitutions are presented.
- Operations should verify that an accepted order can be tracked and that a refund or return reaches the same customer record.
- For an eligible integration pilot, compare completed orders, exceptions and support workload with the team's existing storefront baseline.
Sources
5 references- November 18 readiness checker announcement
Perform.AI via PR Newswire
- Readiness checker and scoring documentation
Perform.AI