Data to decisions in an instant

Before the end of the decade, retail companies that intend to maintain a competitive edge will have to master the art of connecting with a new class of customer comprised of digital agents. Likely to be profound, the implications of the emergence of agentic commerce were top of mind in formal presentations and private conversations at last month’s FMI Midwinter Executive Conference. The reason for the high level of interest is not hard to fathom.

“The rapid progression of generative AI is familiar to everybody in this room,” said Dan O’Connor, executive fellow at the Harvard Business School. “But the commercial understanding of what your business looks like when consumers are managing their own portfolio of AI agents is only just now coming into focus. The core strategic question is what happens as this new way of selling begins to draw business into that model – that is, away from stores, websites and apps.”

O’Connor shared his insights at a panel discussion during a keynote session moderated by Mark Baum, chief collaboration officer and senior vice president of industry relations at FMI – The Food Industry Association. Other panelists included Desirée Gosby, senior vice president of tech strategy and emerging technology at Walmart Global Tech, and Guy Peri, chief information and digital officer at McCormick & Co.

“One of the implications that I think a lot about — and most organizations aren’t yet ready to think about — is that in the world we’ve lived in for the last 20 years, there was lots of time between the data and the decision. The world that we’re heading into collapses that time frame. 

“The competitiveness of agentic commerce is such that data and decisions are going to come together and be taken instantaneously. We don’t know who’s going to influence these decisions. It’s not to say that there isn’t a human involved. It is to say, however, that there’ll be many occasions where consumers will just simply let the agent handle repetitive portions of the things that they buy.”

The development of another tech-driven channel — one in which consumers will have one or more digital agents built into cell phones and other devices — gives retailers and CPG companies reason to enlarge their strategic vision. Walmart is one of the companies that has already done so.

Gosby pointed out that AI has been a component of the business for many years. The retailer has an enterprise-wide agentic framework in place that includes four super agents that give stakeholders — Walmart shoppers and Sam’s Club members, associates, business partners and developers — an efficient means of interacting with the company. 

“We’re already in a world where you’ve got hundreds of apps on your phone,” she said. “We want to reduce the cognitive burden for everyone. Using these agents to understand the problem you’re trying to solve and then orchestrating the response — that helps to simplify things so that teams can then focus on prioritizing what makes sense for them.”

For the consumer, that approach promises to deliver meaningful personalization. By harnessing the power of AI, retailers will finally be able to deliver the individualized shopping experience that has been talked about for years.

“When it comes to things like discovery, the consumer will be interacting with an agent that understands their taste preferences and presents the right products to them in a more personalized way,” explained Gosby. “In an agentic world where you have all these agents helping you, especially with repetitive tasks like grocery shopping, conversion is transformed basically into confirmation.”

The rise of agentic commerce will also test CPG companies. Changes in the manner which they present their brands in the digital realm are under way.

“The big difference with agentic commerce is the autonomy of these agents and how they are going to make decisions on behalf of each of us as consumers,” said McCormick’s Peri. “From a brand perspective, our job is to make sure we get the right information to the right purchase decision, which in this case happens to be the agents. We’re spending a lot of time understanding how do we ensure that attributes that are important in that purchase decision are very clear and that we have the right distribution mechanisms to be able to inform these agentic decisions.

“We are looking at agentic as another channel. A brand’s mission, much like a retailer’s mission, is get the right message to the right consumer at the right time — and that doesn’t change.”

No matter the source of the information about products and services or the target audience — humans or agentic agents — data must be accurate and reliable. 

“These models run on trust and transparency,” said O’Connor. “You must learn how to promote and project data so that it’s discoverable by the LLM [large language model]. You can’t control the LLM; you can control the knowledge graph. It’s the one leverage point that we all have.

“The data that powers the knowledge graph is now your product. I love all sorts of food that you sell, but the data about those products is what you’re selling in the world of agentic commerce.”