What's the fastest way to check if my catalog is agentic-commerce ready?

Work through three questions in order: Discovery (can an agent crawl and find your products?), Completeness (is each product's data. Attributes, specs, price, availability. Rich and structured enough to be shown?), and Checkout (can an agent actually transact. Clear offers, policies, and a compatible checkout?). A gap in any one breaks the chain.

Do I need to support a specific protocol to be ready?

Protocol support (like ACP) is the checkout layer, and it matters, but it's the last mile. Most brands fail earlier, on discovery and completeness: products that aren't crawlable, or product data that's too thin or unstructured for an agent to use. Fix those first; they also improve ordinary AI visibility, protocol or not.

Is your catalog ready for agentic commerce?

An AI agent can only sell what it can find, understand, and buy. That's the whole readiness test, and it breaks cleanly into three questions. Fail any one and the chain snaps.

1. Discovery: can an agent find your products?

If the agent can't reach your product pages, nothing else matters.

AI crawlers are allowed (in robots.txt and at your CDN. A CDN can silently block them, as we've covered).
Products are in server-rendered HTML and structured feeds, not locked behind JavaScript an agent won't run.
Your catalog is exposed where agents look. Clean URLs, a sitemap, and product feeds.

2. Completeness: is the data rich enough to show?

Agents don't see your hero imagery; they read structured product data. Thin data means your product can be found but not confidently recommended.

Product schema (Product / Offer / availability / price) on every PDP.
Full attributes: specs, materials, sizes, categories, identifiers, not just a marketing paragraph.
Accurate taxonomy so the agent can place your product in the right category for a query.
An llms.txt and clean structure so the important facts are easy to extract.

This is where most catalogs fall down: a human shopper fills the gaps from the photos; an agent can't.

3. Checkout: can an agent actually transact?

The last mile: once recommended, can the purchase complete?

Clear, machine-readable offers and pricing.
Published shipping, returns, and policy information an agent can surface.
A checkout the agent can use. Including support for emerging standards like the Agentic Commerce Protocol where you want in on in-chat purchase.

The order matters

Work the three in sequence. Protocol/checkout support is the headline, but it's the last step. Brands lose far more to invisible products and thin data than to missing checkout integration. The encouraging part: discovery and completeness are exactly the same investments that lift your ordinary AI visibility and citations. You're not doing separate work for "AI search" and "agentic commerce". You're making your catalog legible to machines, which pays off on both.

Knowing which of the three is actually holding you back, and tracking it as the engines evolve. Is what an AI-readiness audit, and ongoing monitoring with Buffy Intel, are for.