Training and policies are as important as tech when incorporating AI

When it comes to artificial intelligence, The Vitamin Shoppe CIO Scott Devlin says his company’s strategy is focused on what can “make the business better and improve our capabilities.”

If AI can enhance, upgrade or assist in some way, it’s “something that we’ll go after,” he told the crowd at NRF Nexus. But it’s not quite the same as some other retailers, especially in terms of personalization and customer experience.

“Ultimately, I would love to use it in customer experience, and I think the potential in the future for one-to-one personalization is huge,” he said. However, if an AI tool generated a response to a customer inquiry containing false or misleading information and presented as fact — known as “hallucinations” — it could have serious consequences for The Vitamin Shoppe. “So, for us, it’s more operationally in the background.”

The company uses AI, for example, to be more proactive in bot management for the website, as well as in fraud. It is also testing AI solutions to make writing code more efficient.

Devlin joined Matthieu Houle, CIO, Aldo Group, and Glenn Allison, vice president, IT applications development, Tractor Supply Company, for the conversation. The panel went beyond talk of different technologies and use cases, addressing how they’re managing issues internally.

Christian Beckner, vice president, retail technology and cybersecurity with the National Retail Federation, moderated the discussion. He encouraged those in the room to “engage with NRF in some of the work we’re doing on AI, not only on the business side, but also on the policy side.” NRF is actively involved in a variety of ways, he said.

At Aldo, Houle said, there’s a recognition that “humans are going to make Aldo successful,” with all of their human qualities. There’s an eye toward giving humans “superpowers” through AI, so they can focus on innate skills such as empathy, creativity and critical thinking.

Aldo has several areas of focus, ranging from “AI every day for everyone” (using tools already available such as Microsoft Copilot) to investment and building toward the “holy grail” of demand and sales forecasting, in a program partly funded by the Canadian government.

Tractor Supply, meanwhile, has been rolling out a variety of chainwide initiatives with AI, Allison said, all centered around the customer. The aim here is to provide legendary experiences for customers, helping navigate them toward solutions that best meet their needs in living the “life out here” lifestyle.

That includes AI communicators for store team members; API integration with the consumer mobile app for a seamless experience in areas such as curbside pickup; and generative AI for product information and product recommendations.

Customers might not even know AI is part of the process, Allison said. They just know they’re able to have a frictionless experience, whether in-store, on the app, on the website, picking up curbside or receiving same-day delivery.

“Instead of calling it GenAI … we almost look at it as a ‘genuine AI,’ all centered around the customer experience,” Allison said. “We’re enabling our store team members with the tools to better serve the customer.”

“We’re enabling our store team members with the tools to better serve the customer.”

Glenn Allison, vice president, IT applications development, Tractor Supply Company

In all cases, the panelists noted, training and policies for responsible use are necessary, whether dealing with bias, taking responsibility for “dumb” posts that AI may have written but humans have authorized, transparency, information security, or helping people recognize that data entered into ChatGPT is public.

The Vitamin Shoppe has an IT Prioritization Committee to help establish areas of focus based on more than simply ROI calculations; at Aldo, there’s a small strategic team, but prioritization is done by business leaders in different verticals.

As for the future, Tractor Supply will build out its ChatTSC tool, which will securely answer common questions across the organization, rather than having team members go to the internet or help desk. The company is also looking at ways to handle its ever-increasing computing needs by investing in GPUs, whether local, cloud or hybrid cloud architecture that support the acceleration of AI.

“We want to have managed growth,” Allison said, and to help control costs.

Houle also spoke of the need for companies to “double down” on cleaning up data and changing processes now to be ready for predictive and prescriptive AI. Within a few years, he predicted, that will be a differentiator.