Shopping bots are about to become one more fussy customer standing in the aisle, except this one reads machine-facing files, rejects broken checkout steps, and does not care how pretty the homepage looks. Perform.AI has launched a free UCP and ACP Readiness Checker aimed at online retailers that need to know whether AI agents can find products, purchase them, and follow an order after payment.
I like the practical angle here. There is plenty of fog around “agentic commerce,” a phrase that has been stretched like cheap taffy by vendors eager to slap AI on every sales deck. This tool at least points at a plain operational question: can an automated shopping agent actually transact with a store under the emerging protocols, or will it hit a dead end halfway through?
Protocols Turn Hype Into Storefront Plumbing
The Universal Commerce Protocol, or UCP, arrived in January 2026. The Agentic Commerce Protocol, ACP, dates to September 2025. Both are intended to establish how an AI shopping agent discovers a product, begins and completes checkout, and receives order information afterward. UCP goes further on post-purchase details by carrying the carrier, tracking number, and tracking link.
That sounds dry because it is dry. And dry infrastructure is usually where commerce either works or falls flat on its face. A retailer can have glossy product photography and a clever loyalty program, but an agent cannot buy what it cannot identify, price, place in stock, or route through checkout. Nor will a shopper be thrilled if the agent can order successfully but cannot surface delivery status or return information afterward.
Perform.AI says its checker runs 147 checks across UCP, ACP, and related web standards. The score is out of 100 and applies only to the protocol a store uses. That last detail matters. Scoring a retailer against capabilities it has not adopted would be nonsense, the sort of benchmark theater that makes executives feel busy without telling them what to repair.
Four Questions That Actually Matter
- Can agents find the store? The checker looks for the file that tells agents what the store supports, whether AI crawlers can enter, and whether product prices, availability, delivery terms, and return policies are readable.
- Can an agent buy? It examines checkout from opening the process through completion. For UCP stores, it also checks listed payment methods.
- Can an agent track the order? The post-purchase review considers order status, carrier details, tracking links, delivery dates, refunds, and returns for UCP implementations.
- Can systems connect safely? The tool makes test calls to listed connection points and checks connection guidance, sign-in information, and encryption.
That framework is more sensible than another generic AI maturity score. It centers on the unglamorous handoffs that decide whether an automated purchase can happen without a human stepping in to untangle the mess. Retail technology has never lacked for seductive front-end demos. The wreckage tends to appear in the order-management, logistics, authentication, and returns layers.
A Free Score Is Also a Sharp Lead Generator
There is no need to pretend this is charity. Users enter a store address for a score; a full report requires a name, company, and work email. That is a familiar business-development funnel, and frankly it is a reasonable one. A free diagnostic establishes the problem, then a vendor can offer help fixing it.
Perform.AI says the checker does not require installation or access to a retailer’s systems. It also says the results follow fixed rules rather than using a large language model or an AI agent. For commerce teams, that repeatability may be the best part. Fix a disclosed item, rerun the checker, and see whether the score changed. No black-box chatbot shrugging its way through a compliance-style assessment.
Agentic commerce will not be won by whoever says “AI” loudest; it will be won by merchants whose product, checkout, and fulfillment data hold together when software starts shopping.
Why the Timing Is Worth Watching
Perform.AI is positioning this launch ahead of the holiday peak season, when a broken checkout path is especially expensive. Its broader pitch includes AI Commerce Visibility, announced in October 2025, along with connected checkout, post-purchase, returns, and logistics capabilities. The company says that underlying system handles more than 100 billion parcel updates annually, serves thousands of brands, and includes over 1,100 carrier integrations across more than 160 countries.
Those are sizable operating claims, though the notes do not provide revenue, customer retention, pricing, or independently verified performance data. Investors and retail operators should keep that distinction clear. Scale in parcel updates can be useful evidence of logistics exposure, but it is not automatically proof that protocol-readiness tools will become a durable revenue engine.
Still, this is a credible wedge. If AI agents become meaningful shopping intermediaries, brands will need a way to identify technical gaps before those gaps quietly cost them conversion. Perform.AI has put a ruler on the table. The next question is whether retailers pick it up before the agents start doing the choosing.