Build a Product Catalog AI Agents Can Trust

Holiday shoppers are handing more of their buying to AI. A new 40rty survey of more than 1,600 US online shoppers found that 63% of Black Friday shoppers are likely to use AI this year. Among shoppers already using AI, 2 in 3 have seen it get product information wrong, and 65% of shoppers who got wrong information walked away from the purchase.

Adoption is moving faster than accuracy, and the information gap usually traces back to the product data agents are reading.

To dig into what brands can do about it before BFCM, our VP & Partner Zac Cherin sat down with Tamir Eden, Co-founder of 40rty, the team behind AgentIQ and one of our agentic commerce partners. Here’s what they covered.

Why the Catalog Comes First

When Shopify started talking about agentic commerce, we knew we had to move quickly, both to protect our brands and to figure out what a real service offering looked like in a space few people understood yet. Most of the early tools focused on measurement: treat ChatGPT and other LLMs as a traffic source, track it, and add FAQ content to product pages to improve AI visibility.

That work has value, but it rests on shifting ground. As Zac pointed out, Google's algorithm updates came along fairly infrequently, while LLMs turn over much faster. Reddit was a high-value source for AI visibility until a new model sharply deprioritized it. AEO and GEO strategies built on those signals make it hard to bet confidently.

Tamir's approach starts somewhere more stable. Shopify’s Global Catalog holds every product from Shopify's merchants in one place, and it's what Shopify distributes to ChatGPT and what agents read when they search. It's grounded in product data, which makes it measurable and comparable over time. “There's grounded truth somewhere,” Tamir said, “and the catalog is the grounded truth.”

For Zac, that idea reshaped how we work. “Taking the catalog-first approach has become our whole philosophy around agentic commerce,” he said. Shopify built the Catalog to solve the problem every brand is worried about, which is getting ChatGPT and other LLMs to recommend their products. The most useful work a brand can do is improve its own catalog while Shopify continues to develop the global one.

Where Shopify Is Taking It

Shopify keeps adding infrastructure on top of the Catalog. Brands can now publish Agentic Product Listings, which only agents see and which sit separately from the listings shoppers browse on the website. Tamir described this as what everyone had wanted: the freedom to give agents detailed, structured data while keeping the storefront written for humans.

Shopify's agentic sales channel also included Shop, which drew some questions when it launched. The connection is now clearer. 40rty's research found an 85% match between Shopify Catalog and Shop results, and the recent release of the Shop SDK for agents opens Shop to any agent that wants to transact. With personal agents like Muse and Instinct gaining traction, Shop is positioned to be the rails agents use to buy. Every improvement a brand makes to its product data today carries forward to a future where a personal agent finds the best product for its user and buys it on the spot.

Your Pre-BFCM Agentic Checklist

Most brands head into code freeze about now, which limits changes to PDPs, merchandising, and email flows. Tamir pointed out that agentic work is the exception. Everything below runs through product data and the agentic channel, outside the theme your shoppers see, so there's still real room to optimize in the weeks before the holidays.

1. Ask an agent to buy from your store

Open ChatGPT, Muse, or Instinct and try to buy one of your own products. It sounds simple, but Tamir noted that plenty of brands haven't tried it yet, and it shows you exactly what an agent-assisted shopper experiences.

2. Ask how agents describe your brand

Ask different agents about your warranty, returns, delivery, and reviews. Tamir shared an example from furniture brands 40rty works with, where agents were asked what happens if an item arrives damaged. The answers reveal whether agents are reading your policies and how much assurance they have in recommending you.

3. Fix taxonomy and metafields

Run a Catalog Audit in AgentIQ to get your Catalog Health score, a baseline on how complete your product data is, even across very large catalogs. From there, fill the category and metafield gaps it surfaces. Your theme doesn't depend on these fields, so they're safe to update during code freeze.

4. Test prompts and publish agentic listings

AgentIQ simulates real shopper prompts and personas against the Shopify Catalog and shows where your listings rank. From there, you can publish Agentic Product Listings built for how agents search.

For most brands, agentic is still a single-digit share of revenue. Tamir sees it as the biggest new opportunity of this BFCM regardless, since the channel barely existed a year ago.

Taxonomy, Explained

Taxonomy is one of the harder concepts to explain to brands, so Zac asked Tamir for a quick primer.

Shopify product types used to be a free-text field. Merchants could enter anything, typos included, which made products hard to index. Shopify responded with the Shopify Standard Product Taxonomy, an open-source tree of product categories that Shopify updates every three months. Each category carries its own set of attributes, and those attributes become metafields on your products.

The deeper the category, the more an agent can learn about the product. A T-shirt filed under Clothing picks up attributes like fabric, size, and care instructions, but nothing that tells an agent whether it's a crew neck or a V-neck. File the same shirt under T-Shirts, and it inherits attributes like neckline, sleeve length type, and top fit, 15 in total. Filling in those fields is what helps a product surface accurately when an agent searches for something specific.

Tamir called this the most important part of product hygiene right now. When agentic search started, 40rty noticed that products with taxonomy and metafields in place were the first to surface for more detailed prompts. The challenge is the labor, because closing the gaps in Shopify means working through products one by one. AgentIQ's Autopilot handles that at scale, choosing the right category for each product and filling the fields that come with it, inside the guardrails your brand sets.

From Discoverable to Chosen

A complete catalog makes a brand discoverable. Getting chosen takes ongoing work. Zac noted that a Catalog Health score moves in both directions: brands add products and restructure data for their own business reasons, while Shopify and 40rty continue learning what drives rank in the Catalog. A one-time cleanup is a strong start, and the gains hold when the work continues.

AgentIQ runs that as an optimization loop. Audit the catalog, set Guardrails for what good data looks like for your brand, close the gaps, and publish Agentic Product Listings. Then confirm each listing ranks in the Catalog and track whether it drives sessions and orders over the following 30 to 60 days. Shopify's agentic channel now reports conversion directly in the admin, and AgentIQ breaks it down further by agent, product page, and collection.

Shopper prompts are getting richer, too. Tamir's example was someone looking for the best running shoes for a spouse who just started a running course, bought as a birthday gift. Dozens of skincare brands may compete for the same holiday gift prompt, each selling to a different persona and use case. AgentIQ works to carry that brand context into the listing itself, where agents can read it.

Building Confidence for Agents

Later in the conversation, Tamir offered the framework that ties it together. A human shopper might click ten links and browse a few pages of products before deciding. An agent sends hundreds of queries to the Shopify Catalog and matches the results against everything it knows about its user, from size to preferences. The clearer your product data is about fabrics, materials, warranties, and anything else an agent might weigh, the more confident it becomes in recommending you.

Tamir boiled it down to one question: ‘How do you build confidence for agents?” That exact framework has become the lens we use with our own brands.

Tamir expects some brands to reach a double-digit share of agentic revenue as soon as next year. On our side, more non-Shopify merchants are coming to migration conversations specifically because Shopify is ahead on agentic. Once they're on Shopify, the catalog is where the opportunity starts.

Watch the Full Conversation

The full video covers more than we could fit here, including how 40rty thinks about agentic analytics and why focused partners matter in a space moving this quickly.

Take me to the video

To get your catalog ready before BFCM, visit 40rty to install AgentIQ and see your Catalog Health score, or reach out to our team to learn more about how we can support your brand’s agentic commerce strategy.

Talk to our team