Product Data Is the Source the Agents Read with Ben Riggle of Channable

Isaac Morey sat down with Ben Riggle at Shoptalk Fall in Nashville, and Ben started with what Channable does for a retailer before he got anywhere near the buzzwords. He is the managing director for Channable in North America, and the conversation kept coming back to one asset, the product data.

“I’m the managing director for Channable here in North America. And what Channable does is it takes all of your product data and makes sure it’s optimized for every surface that you want to go out on as well,” he says.

Agentic Discovery and Agentic Checkout

Isaac asked for the quick rundown on agentic commerce, since some people he asks have a million things to say and others turn it around and ask him what it is. Ben split the term into two parts.

“to me agentic discovery is what we’re all doing every day in our lives. We’re relying on AI mode, Google, et cetera,” he says.

Checkout is the other part.

“agentic checkout is a little bit of a different beast,” he says.

Ben also said the word may not stay in use for long, because the surfaces are blending together.

“I think agentic is a great term for now, but I do think we’ll just start to see this as just kind of search and discovery and kind of across all these surfaces,” he says.

Isaac compared it to buying on Amazon without thinking about the platform, and Ben brought in his mother.

“I don’t think my mom thinks about it that way. She just, you know, sends something to her house, it gets there and she’s happy however it gets there,” he says.

Product Data Is the Source the Agents Read

Isaac asked how a merchant who tracks traffic from Google can tell whether visitors are arriving through ChatGPT or Claude. Ben said the platforms are still working that out with merchants, and he named what he does know.

“What we do know is that when you think about product data, that is the source that these agents and these LLMs are relying on to make sure they’re providing an accurate recommendation,” he says.

He described what better data looks like to the platforms Channable works with.

“the better your product data is, the more hygienic it is, the better fill rate you have on all of the different attributes, the more likely you are to surface in a conversation,” he says.

Proving the link is another matter.

“the attribution for that is gonna be messy for a long time,” he says.

Hygienic Data and the Attributes Google Added

Isaac asked what hygienic product data means. Ben described the required, nearly required and optional attributes every platform has, and then what changed once large language models entered the picture.

“With an LLM now, Google’s added for example 30 conversational attributes,” he says.

Those attributes are harder to fill than a title or a price.

“and so now you have things like Q&A pairings or you have reviews or you have rankings and these things are a little bit more abstract as well,” he says.

AI Talking to AI

Isaac asked whether it is acceptable to use AI to answer the AI’s own question, such as the Q&A on a product page. Ben said the panel he had just left asked the same thing, and he used a movie to describe the risk.

“It’s like it’s AI talking to AI. It’s like the multiplicity movie. I’m probably dating myself by that where you have three different versions of Michael Keaton and each one gets a little bit crazier,” he says.

His answer for Channable is rules wherever a rule will do the job.

“a lot of what you’re doing here is deterministic rules-based, so we can find something that’s happening on one or two or three different products,” he says.

“And we can fix that with rules rather than having an AI or an LLM come in and change everything in your product set as well,” he says.

Isaac also asked every guest at the table to define AI slop, and Ben gave his definition.

“I think AI slop is stuff that’s written by AI and read by AI and then regurgitated by people without understanding, right?” he says.

He tied the idea back to data inside a company.

“you’re gonna have a lot of different conclusions driven off a lot of different data sets,” he says.

The AI Readiness Tracker

Isaac asked about new features, and Ben described the tool Channable launched.

“we’ve just actually launched an AI readiness tracker within Channable,” he says.

The tracker starts with a feed the merchant already has.

“you just attach your Shopify or your Google Merchant feed and it will actually score and tell you, you know, how ready is your feed for AI,” he says.

The Hot Take on Data Control

Isaac closed by asking for a hot take. Ben said the complexity people picture around AI shrinks when he looks at what companies are doing.

“it’s a lot of the same things that made companies successful 10, 15 years ago, which was, you know, having really tight control on their data, getting federated access to that data to small groups, and then allowing that to be democratized among their teams as well,” he says.

“I think AI is just how you use that data now,” he says.

He also had a word for the businesses that feel behind.

“you’re probably not as far behind as everybody out on the show floor or on the stages would make you think you are,” he says.

Channable is at channable.com.