How Catalog is helping retailers show up on ChatGPT

If your product data is a mess, chances are your brand won’t show up well on AI platforms, including ChatGPT.

Catalog, a startup co-founded in April 2025 by fintech and commerce veterans Hamish Gunasekara and Dylan Farrell, is trying to solve that problem for retailers. Based in San Francisco, Catalog works with about half a dozen brands to clean up their product data—including title, price, and descriptions—so their items can show up when people search across AI platforms.

“Catalog is building critical infrastructure for how commerce will work in an AI-first world,” Lauren Kolodny, general partner at Acrew Capital, wrote in a March release announcing the firm’s lead in Catalog’s $3 million pre-seed round. “As AI agents take on a larger role in discovery and purchasing, structured and reliable product data becomes essential. Catalog ensures merchants can show up and compete in that environment.”

The company is all set to sign on two big enterprise clients—one in the personal care space and a department store chain—in the coming months, Catalog CEO Gunasekara told Retail Brew.

The idea for Catalog came out of everything Gunasekara and Farrell picked up in their years working across fintech, commerce, and AI product systems. Gunasekara was an early data scientist at buy now, pay later platform Afterpay (which Jack Dorsey’s Block acquired in 2022) and later led merchant integrations across Square and Cash App. Farrell studied mathematics at Harvard before spending years building product recommendation engines at fintech firm Curinos. Given their combined experience, Catalog seemed like the natural next step.

The method behind the madness: Catalog takes all of a retailer’s unstructured product data, runs more than 50 LLM calls (one LLM call is essentially a prompt sent to an AI model) to clean and transform this data and make it more palatable for AI agents in the form of JSON (the language that AI is most comfortable in).

On the technical side, the AI agents read just raw text in the JSON format— “a black screen with white text,” Gunasekara said. This process ultimately results in a structured product data set that’s more likely to be picked up in search results when someone is looking for items on ChatGPT, Gemini, or any other AI platform.

“We want to be helping a brand whenever AI is trying to interact with their products,” Gunasekara said. “We effectively are a third store for them. You have in-store, online, and then you have agentic.”

Catalog, so far, is working with smaller, niche brands such as Hobbiesville, Kidsy, Splits59, and Feals. Recently, baby and kids discounter Kidsy went from being invisible on AI to generating more than 2,100 shoppers directly from AI assistants, Gunasekara shared via email.

For Gunasekara, commerce is changing, which makes online shopping a fun problem to solve.

“You flash back to 12 months ago—brands were not thinking about AI as a shopping channel, they were focused on things like tariffs and just other stuff,” Gunasekara said, adding that retailers’ perspectives on AI became more favorable around Black Friday 2025. Currently, he said, about 98.5% of AI traffic to brands is coming from ChatGPT.

Gunasekara recalled that co-founder Farrell realized during his junior year at Harvard that AI is only as good as the data infrastructure underlying it, and that’s where he wanted to focus.

At Afterpay, Farrell and Gunasekara built the Afterpay Shop app, an early version of what you’d now call agentic commerce.

“We had to work with product data there and broadly, I consider my experience as always being at the frontier of how people shop, creating new experiences for how people shop,” Gunasekara said. “Our 20-year vision is to become the operating system for agentic commerce…The wedge is really great product data sent to the AI systems that very naturally goes into checkout, into payments.”