The emergence of intelligent, agentic artificial intelligence assistants capable of automating purchasing decisions has triggered a profound strategic confrontation within the retail sector. As consumers increasingly delegate shopping tasks to these conversational agents, major retailers face the existential risk of losing direct control over the core elements of their business: how their products are displayed, recommended, bundled, and sold. This deep-seated concern is pushing large chains, including grocery giant Kroger and home improvement retailer Lowe’s, to pursue a sophisticated, hybrid AI strategy aimed at retaining proprietary control and customer data, even while leveraging the powerful foundational models developed by companies like Google and OpenAI.

The fundamental shift presented by agentic commerce is the potential loss of the digital ‘shelf space.’ For decades, retailers invested heavily in store layout, website search optimization, and retail media networks to secure visibility and influence product discovery. The rise of AI agents, which can fulfill complex customer requests with a single instruction—such as "order ingredients for a low-carb family dinner" or "find tools for an intermediate tile installation"—moves the competitive battleground from the store aisle or website to one of "algorithmic preference." Once a customer’s goal is handed over to a third-party, general-purpose AI, the retailer risks becoming a fungible fulfillment partner, losing the opportunity for high-margin add-on sales, weakening hard-won brand loyalty, and seeing valuable advertising revenue migrate to the platforms that control the conversation. This explains why some retailers are approaching partnerships cautiously, choosing to embed AI within their own proprietary ecosystems rather than simply selling through external chatbots.

Kroger and Lowe's are actively navigating this complex terrain by adopting a multi-vendor approach, yet insisting on a custom layer of control. Kroger, for example, is utilizing a variety of agentic tools. While the company has expanded its relationship with Google Cloud to deploy the Gemini Enterprise for Customer Experience (CX) solution, including a new Shopping assistant, its primary mechanism for control is through data integration. The retailer’s goal is to ensure the AI's recommendations are "grounded in actual assortment, pricing, and availability," and are based on Kroger’s extensive proprietary data asset, including customer price sensitivity and brand preferences. This strategy keeps critical decision-making logic—the bundling and offering of its products—within the retailer's own systems, rather than allowing an external platform to make purchasing decisions solely based on general model data. Separately, Kroger is also an early adopter of Instacart's AI assistant, Cart Assistant, available within its mobile app, signaling a calculated strategy to spread its technology risk and capabilities across multiple, specialized providers.[1][2][3][4]

Lowe's has implemented a similar, even more explicitly hybrid model. The company's customer-facing virtual assistant, Mylow, which has been shown to more than double conversion rates for online shoppers who use it, is powered in part by a partnership with OpenAI’s ChatGPT. This use of a leading large language model is paired with an underlying architecture that manages a variety of vendors. As a major enterprise, Lowe's does not rely on a single provider, working with several, including OpenAI, to avoid platform lock-in. Their internal strategy is to build a centralized platform that can abstract the engineers and developers from the specifics of rapidly evolving models and protocols, thereby maintaining agility and control over the business logic and customer experience. This allows the home improvement retailer to integrate complex functions, from home visualization to guiding an associate through a plumbing repair project, while the retailer retains ownership of the customer interaction and the data generated from it.[4][5][6][7]

This pursuit of domain-specific control by retailers has significant implications for the wider AI industry. The movement signals a major shift away from a monolithic AI provider model toward an ecosystem focused on specialized, agentic platforms. The value is no longer just in the general-purpose foundational model, but in the retailer's ability to layer its unique business data—such as its supply chain metrics, local inventory, private-label assortment, and customer loyalty profiles—on top of that model. This has created a new operational imperative for enterprise retailers: to develop a dynamic, data-centric architecture capable of ingesting live inputs and recalibrating AI models on demand. Instead of simply generating insights for human interpretation, the new generation of agentic systems must be capable of executing decisions, such as dynamic pricing adjustments, real-time inventory rebalancing, and personalized product recommendations, all while maintaining a “human-in-the-loop” control mechanism to ensure accuracy, compliance, and brand consistency.[8][9]

The strategic efforts of companies like Kroger and Lowe's illustrate that the next competitive arms race in commerce will not be won by the company with the most powerful general AI, but by the retailers best able to leverage their proprietary domain knowledge. By adopting an agentic approach that prioritizes data control and a multi-vendor technology stack, retailers are fighting to prevent the customer experience and the subsequent economic value from being entirely intermediated by a handful of giant technology corporations. Their challenge is to move from being merely 'AI-enabled' to becoming 'agent-preferred'—the brand that algorithms consistently favor because of its precise, up-to-date, and controlled product knowledge. This necessity for control is not just a technological choice, but a fundamental business strategy for preserving market share and brand identity in the age of automated, conversational commerce.[10]