Retailers racing to embrace agentic AI are confronting a strategic choice that may define their competitive relevance for the next decade: whether to plug into massive third-party AI platforms for instant scale, or invest in proprietary agents to retain control over their customer relationships, data and brand identity.
Tom Sebok, managing partner at the New England Consulting Group, added that most large retailers are no longer debating whether or not agentic AI will reshape commerce, but are considering how it will do so, and on whose terms.
“This has exploded over the last 24 months,” Sebok said, noting that the technology itself has advanced exponentially. Major retailers, he added, are fully aware of the opportunity. The harder question is where to place long-term bets: speed versus ownership, partnership versus independence.
At the heart of that decision is what Sebok describes as a “sovereignty trade-off.” Partnering with platforms such as ChatGPT or Gemini offers immediate reach and technical sophistication, but risks relegating retailers to the role of back-end fulfillment providers. Building proprietary agents, by contrast, preserves customer sovereignty but requires time, capital and technical depth.
“The risk for all retailers,” Sebok said, “is allowing themselves to become a back-end fulfillment center for somebody else’s AI solution.”
Compounding that challenge is what Sebok calls the “collapse of the funnel.” In an agentic AI world, the traditional shopping journey — discovery, research, comparison and checkout — is compressed into a single interaction. The AI agent identifies intent, makes recommendations and executes transactions in one step.
“That’s not just a new search bar,” he said. “It’s the collapse of the shopping journey itself.”
In that environment, data becomes the primary currency of relevance. Retailers that cede customer data to third-party agents, Sebok argued, are effectively paying “digital rent” on their own shoppers. The winners will be those that use AI to deepen their direct data moats rather than drain them.
Yet even strong data ownership is no longer sufficient. Agentic AI is rewriting the rules of visibility, particularly in marketing. Traditional SEO — long driven by retailer-controlled websites and published content — is losing influence as AI agents rely more heavily on broader public discourse.
“These tools are far less reliant on what you put on your website,” Sebok said. “They’re far more influenced by what the rest of the internet is saying about you — Reddit, ratings, reviews, social commentary.”
That shift forces brands and retailers alike to rethink marketing as narrative management across the entire internet, not just owned channels. Consistency, clarity and disciplined messaging become essential as AI systems ingest unstructured, uncontrollable conversations to generate recommendations.
Against this backdrop, Sebok described one recent development as “seismically game changing”: Apple’s decision to integrate Google’s Gemini AI into its ecosystem. The partnership, he said, combines the world’s most ubiquitous consumer hardware with the world’s deepest behavioral data pool, creating an AI retail enabler with unprecedented reach.
“If I’m a retailer,” Sebok said, “figuring out how to make friends with this Apple-Google partnership is priority No. 1.”
And on the question of whether retailers should build their own ecosystems or integrate with the dominant AI platforms, Sebok suggests that most companies will need to do both.
“You need to serve the customers you know better than anyone else,” he said, “but you also have to be preferred in a world you don’t control.”
The parallel, Sebok noted, is foodservice: Brands like Domino’s and McDonald’s built powerful owned digital ecosystems while simultaneously participating in delivery marketplaces they didn’t control.
For retailers, agentic AI raises the same existential question. Can they remain trusted brands with authority and expertise, or will they become interchangeable utilities in someone else’s network?
In an AI-mediated marketplace, Sebok suggested, that distinction may be the difference between relevance and invisibility.
