NRF 2026: From Shopping to Store, from Store to Supply Chain — the Age of Assistants
NRF 2026, held at New York’s Javits Center, continued to send strong signals about the future of retail technologies. As expected, the main theme of this year’s conference and expo was artificial intelligence—but unlike previous years, we are now seeing AI embedded in real, everyday use cases. Today, 15% of retail IT budgets are allocated to AI, and this share is growing at an annual rate of 27%.
Industry briefings revealed that AI investments are moving beyond pilot projects and limited chatbot deployments. Instead, they are taking shape within a three-layer investment framework that simultaneously transforms both the customer-facing front end and the operational “back of house.” On the front end, personalized recommendations, dynamic pages, and customer assistant scenarios that research and compare on behalf of the shopper stand out. In the background, decision-making mechanisms—such as demand forecasting, inventory and supply chain management, pricing and promotion optimization, and workforce planning—are being rearchitected with AI.
Enabling all of this is an invisible foundational layer that includes AI infrastructure usage costs, data platforms, security and governance, and data anonymization. In short, building a sustainable, AI-ready infrastructure has become the baseline requirement.
The Rise of Agent-Based Commerce
The most prominent AI-related message at NRF—one that directly addresses future customer behavior—was the shift from AI-assisted shopping toward “agent-based commerce,” where autonomous agents can make decisions on behalf of users. Several major announcements were made to support this transition.
Last year, solutions presented as “agents” were essentially advanced chatbots. This year, however, systems capable of true autonomous decision-making, transaction execution, and managing cascading impacts were showcased.
Protocol Wars: UCP vs. ACP
One of the most critical announcements of NRF 2026 came from Google: the Universal Commerce Protocol (UCP). Introduced in partnership with Google, Shopify, Etsy, Wayfair, Target, and Walmart, UCP is positioned as “the infrastructure for commerce within AI interfaces.” Its biggest advantage for retailers is that companies already integrated with Google Merchant Center are automatically included. In other words, if you are feeding your product data to Google, Gemini can already find you.
On the other hand, Agentic Commerce Protocol (ACP)—previously announced by OpenAI and Stripe—offers a different value proposition: for retailers using Stripe, a single line of code enables visibility directly inside ChatGPT. While ACP focuses on speed and simplicity, UCP stands out with features such as user authentication and loyalty program integration.
These two protocols can be thought of as the HTTP of agent-based commerce. The answer to which one will win may well be “both.” Walmart’s strategy makes this clear: by adopting both ACP and UCP, the company operates under the motto of “being present wherever customer intent begins.” This signals a future that goes beyond omnichannel toward “omni-protocol” commerce.
It is also worth noting Amazon’s long-standing position. Amazon effectively says, “I am already my own standard—through Rufus, you can discover and order both products in my ecosystem and those outside of it.” Some in the industry frame this as a battle between Walmart-led open standards and Amazon’s closed network approach.
Agent-Based Commerce: Redefining the Consumer Experience
One of the most debated sessions at NRF was AWS’s David Dor presenting the “Agentic Commerce Maturity Model.” Spanning from Level 0 (assistant-based recommendations) to Level 5 (fully autonomous shopping agents), the framework clearly illustrates where the industry is headed.
Today, most retailers operate at Level 2–3, experimenting with guided and transactional use cases. However, expectations are set for a transition to Level 4 and 5 by 2026–2027, where AI agents make purchasing decisions entirely on behalf of consumers.
Adobe’s data puts this transformation into perspective: during the 2025 U.S. holiday season, AI-driven e-commerce traffic increased by 693% year over year. Salesforce reported that AI and agents accounted for 20% of holiday retail sales in 2025, reinforcing findings that AI adoption is strongest in customer service-related processes.
Terminology is also evolving. Alongside B2B and B2C, A2A (Agent-to-Agent) commerce is now emerging—where a consumer agent communicates directly with a retailer’s agent. Managing not only customers but also agents is becoming a core operational topic.
AI Reflections from Retail Giants
Amazon’s approach stands out as notably different. With its AI assistant Rufus, which reaches 250 million customers and has already generated $10 billion in sales, Amazon is choosing to strengthen its own ecosystem rather than join UCP or ACP. CEO Andy Jassy’s statement—“We will reconsider when other LLMs deliver a compelling experience”—reflects strong confidence in their position.
Target, meanwhile, is pursuing a different path. By developing a custom application inside ChatGPT, customers can use the “@Target” command to search products, check availability, and complete orders. This API-based integration model operates independently of protocols and also enables Target to understand customer intent—even to the point of deciding whether to add missing products to its assortment.
Another notable development is Kroger’s partnership with Instacart. By transforming the in-store experience with smart shopping carts, the collaboration delivers real-time basket total visibility, driving an average 2–3% increase in sales. Deployed in 20% of Wegmans stores, the system also boasts zero shopping cart theft, an impressive operational outcome.
Walmart’s strategy of “being where customer intent begins” serves as a manifesto for the industry. The challenge is no longer simply being where the customer is, but being present at the exact moment intent is formed—and that intent is increasingly shaped within AI assistants.
The New User Interface: Voice and Visual Recognition
Solutions showcased at NRF 2026 made it clear that user interfaces are undergoing a fundamental transformation. Several examples demonstrated voice-based interactions within stores, enabling private, targeted communication between kiosks and customers. Shoppers can verbally express their needs, and systems can recommend appropriate products. Specialized speaker and microphone designs allow these interactions to occur without echo or noise disruption in the store environment.
On the visual recognition front, on-device camera systems operating without cloud connectivity stood out. Using a single camera, retailers can perform loss prevention, self-checkout verification, employee misconduct detection, and customer behavior analysis.
In the realm of physical AI, the convergence of robotics, increasingly affordable RFID, and visual recognition technologies is driving significant improvements in inventory accuracy across hundreds of stores.
What It Means to Be AI-Ready: Problem + Data + Process
The most meaningful takeaway from NRF 2026 is that it is no longer enough for AI to “be capable.” Organizations must first answer the question: “Should it be used?” Extracting value from AI rests on three pillars.
First, a clearly defined business problem—not “let’s use AI,” but “let’s increase inventory accuracy to 98%.”
Second, high-quality, consistent, and accessible data to fuel the solution—no matter how intelligent agents become, their value is limited by the data they consume.
Third, mature business processes where data is produced and decisions are executed.
Dollar Tree provided a compelling example. In early experiments, they designed well-intentioned but unrealistic processes, such as asking chatbots for guidance during unexpected crisis scenarios—approaches employees did not trust. They later pivoted to a more meaningful use case: an AI-driven operational decision support system that helps store managers identify which store to focus on, which KPIs matter most, and where risks and opportunities lie. This shift delivered tangible value.
Another example: when an order gets stuck, an AI system detects the issue, evaluates possible solutions, selects the optimal one, and manages downstream impacts with minimal human intervention. However, this level of autonomy requires clearly defined processes and business rules upfront. No AI can make sound decisions in an environment where rules are ambiguous.
Only when problem, data, and process are addressed together do AI investments create value. Otherwise, technology remains a cost center. Companies that have invested for years in data quality, process standardization, and system integration are entering this transformation with a built-in competitive advantage.
The Core Message from NRF 2026
The key message from NRF 2026 is clear: the era of agent-based commerce has begun. The companies that will move fastest are those that have long invested in data quality, process standardization, and system integration. AI is becoming accessible to everyone—but those who use it effectively will be the ones who are prepared.
The winners of the digital commerce protocol wars have yet to be determined. But the winner of the data war is already clear:
Those who own the data will own the future.