An audit tool, not an enabler.

The Agentic Commerce Protocol shipped September 29, 2025. Universal Commerce Protocol shipped January 11, 2026. In the seven months between, six dated ACP releases and two dotted UCP revisions changed what counts as a ready store. Most merchants we talked to in early 2026 — on Shopify Community, BigCommerce forums, WooCommerce Slacks — were running audits from blog posts written six weeks earlier with no way to tell what was current.

Shopify shipped their own free tool in April 2026. It's a fast single-product-page schema check — but it won't score a Shopify store at all (it just says "you're set"), and for BigCommerce or WooCommerce it returns a one-page score next to a "switch to Shopify" pitch. It's one-shot, single-page, and doesn't tell you when the spec changes. BigCommerce, WooCommerce, and the agencies running mixed-platform rosters had no neutral, whole-store equivalent.

We built AI Commerce Audit to fill that gap. 82 checks across all five AI shopping surfaces, every audit stamped with the spec versions in effect that day, and a public methodology so any technical buyer can verify what we're measuring.

• The score should always be free. You need a number you can share — on social, in a deck, with a client — without buying anything first.
• Honest comparison beats marketing. If a competitor is the right answer for a given buyer, we say so — see /platforms/shopify.
• Public methodology is the point. Our scoring math, severity weights, surface weights, and pinned spec versions are public. A buyer who can't verify how we score shouldn't trust the score.
• Polite crawlers are non-negotiable. Single connection per host, robots.txt honored, identifiable UA. Always. See /bots.
• We won't score things we can't measure. Microsoft Copilot's merchant docs are thin; we mark Copilot-related checks as inferred and don't pretend to know more than the spec does.

AI Commerce Audit is built and operated by Vlad Duma. Vlad's background is in ecommerce platform integration and B2B SaaS engineering. We publish our scoring methodology and the exact spec versions we pin to, so any technical buyer can verify what we measure.

We're not VC-backed. We're not running an ICO. We charge for the parts of the product that take ongoing engineering work (monitoring, score history, and multi-store tracking) and keep the audit itself free.