Can AI actually complete the recommendation — specs, price, stock, where to buy?
Getting named is half the sale. The other half is whether the model can find the cotton pants in your catalog, quote today's price, confirm stock — and send the shopper somewhere that actually ships to them. Product Readiness makes your catalog completable, end to end.
The sale that died at the finish line
An AI shopping assistant called [redacted] recommended a black designer bag from [redacted]. The shopper said yes — and was routed to a boutique that doesn't ship to her country. No fallback, no second link. No sale.
Brand names redacted for legal reasons. The transaction — and the lost revenue — are real.
This failure is invisible to everyone: the model doesn't know, the shopper blames the brand, and the brand never finds out it happened. Multiply it by every answer that ends in a marketplace link, a dead region, or a stale price, and that's the gap between being recommended and being bought.
Legible: the model knows you — does it know your products?
Models learn brands from the open web; they learn catalogs from structured data or not at all. Connect your Google Merchant feed or Shopify store once, and Herm keeps an agent-readable mirror of every SKU — specs, price, stock — synced continuously, so the answer at 9:41 reflects your catalog at 9:41, not a crawl from March.
One feed, every protocol
There is no standard for agent commerce yet — there's a race. ACP wires products and checkout into ChatGPT; UCP wants to be the common commerce language across Google's surfaces and a wider ecosystem; MCP exposes catalogs as tools an agent can call — and the list keeps growing, renaming, and merging. None of them has won. Your feed shouldn't have to care: connect once, and Herm keeps your catalog conformant as the standards move — the same verified data, spoken in every protocol that matters, updated by us so it isn't a project for you.
Closeable: a where-to-buy that ships
Every destination in your feed carries ships-to and region eligibility, and you rank the routing: your own store first, marketplace fallback where you allow it, region-eligible retailers where you don't ship direct. The London shopper gets the London answer. The margin — and the customer relationship — route where you decided, not where the crawler landed.
Audited like an agent shops
The Deep Audit walks your catalog the way an agent does — Find → Answer → Trust → Close — and scores each stage. Marketers see the plain-language issue and its impact; your developer gets a precise work-order export. Pass the bar and your feed carries the Herm Listed / Herm Verified badge — the mark that tells the source layer your data can be trusted.
Product Readiness is rung two. Found gets you named; sellable gets you quoted with a live price and a link that closes.
See the whole ladder →Questions
What do I connect?
A Google Merchant feed URL or the Shopify app — one connection, continuous sync. Bespoke stacks are scoped in early access; we'd rather be honest about an adapter than promise universally.
What does the audit actually check?
Four agent-journey stages — can an agent find the product, answer specs and price, trust the data, and close with a valid, region-eligible destination — with named issues per SKU and a developer-ready work-order.
Which agentic-commerce protocols do you support?
The moving set — ACP, UCP, MCP — with new ones tracked as they emerge. You connect one feed; conformance per protocol is our job, continuously, not a migration project for yours.
Can I keep shoppers off marketplaces?
You rank the destinations. Own store first, marketplace as fallback or not at all, region-eligible retailers where you don't ship direct — the routing is your policy, enforced in the data.
Does a verified feed guarantee I get recommended?
No — distrust anyone who says otherwise. It guarantees you're completable: findable, quotable, and closeable. Whether you're chosen is what rungs one and three are for.
Make every recommendation completable.
Join early access and get your catalog audited with the founding cohort.