NRF 2026 Recap: Every Agentic Commerce Announcement That Matters
NRF 2026 -- Retail's Big Show -- took over the Javits Center in January with a single, unmistakable message: agentic commerce is no longer a concept deck. It is shipping code. Over the course of three days, Google open-sourced a new commerce protocol, Stripe and OpenAI launched a rival standard with
NRF 2026 Recap: Every Agentic Commerce Announcement That Matters
Last updated: March 2026
NRF 2026 – Retail’s Big Show – took over the Javits Center in January with a single, unmistakable message: agentic commerce is no longer a concept deck. It is shipping code. Over the course of three days, Google open-sourced a new commerce protocol, Stripe and OpenAI launched a rival standard with 25+ endorsers, Microsoft turned its Copilot chat window into a checkout lane, and Target committed to letting shoppers buy groceries inside ChatGPT. For the first time in NRF’s history, the conference stood up a dedicated AI Stage, and the phrase “agentic commerce” appeared in more session titles than “omnichannel.”
This article catalogues every major announcement, the protocols and partnerships behind them, the stats that frame the opportunity, and the concerns retailers voiced behind closed doors. If you left New York early – or skipped the trip entirely – this is everything you need to know.
Google: The Universal Commerce Protocol and Native Checkout
Google CEO Sundar Pichai delivered the keynote on January 11, unveiling the Universal Commerce Protocol (UCP) – an open-source standard designed to let AI agents discover, evaluate, and purchase products on behalf of consumers through a single integration point.
By February 2026, UCP-powered checkout was live for U.S. shoppers purchasing from Etsy and Wayfair, with Shopify, Target, and Walmart integrations announced as coming soon.
Crucially, UCP is designed to interoperate with every other major agent standard: Anthropic’s Model Context Protocol (MCP), Google’s own Agent2Agent (A2A) protocol, and Stripe/OpenAI’s Agentic Commerce Protocol (ACP). The specification is published on GitHub, with Python SDKs and reference server implementations available for developers.
Google also announced several adjacent capabilities:
- Business Agent – an AI-powered sales associate embedded in Google Search, launched with Lowe’s, Michael’s, Poshmark, and Reebok. Retailers can train the agent on their catalog and brand voice through Merchant Center.
- Gemini Enterprise for Customer Experience – a Google Cloud platform (now in preview with Kroger, Lowe’s, and Woolworths) that extends agentic AI from the shopping interface into post-purchase operations: returns, exchanges, and customer support.
- Agentic Price Tracking – users set a target price in Gemini, and the agent monitors the Shopping Graph, notifies when the threshold is hit, and can auto-checkout via Google Pay on the user’s behalf.
One stat that circulated widely on the show floor: AI search orders on Shopify stores are up 11x since January 2025.
Stripe and OpenAI: The Agentic Commerce Protocol and Shared Payment Tokens
On the payments side, Stripe and OpenAI jointly introduced the Agentic Commerce Protocol (ACP) – an open specification for embedding commerce into AI applications. ACP launched with 25+ endorsing partners, including Salesforce, Squarespace, PwC, and commercetools. The full specification is published on GitHub.
The marquee technical innovation is Shared Payment Tokens (SPTs) – a new payment primitive that lets AI agents initiate transactions using a buyer’s stored payment method and explicit permission, without ever exposing card credentials to the agent or the merchant. SPTs address the trust gap that has kept autonomous purchasing theoretical: the consumer grants scoped authorization, the agent executes, and Stripe settles.
Stripe also unveiled its Agentic Commerce Suite, a single integration point that automatically supports all major agentic protocols – ACP, UCP, and others – through one setup. For merchants already on Stripe, this means protocol compatibility without incremental engineering. commercetools announced early enterprise adoption of the suite at the show, and JD Sports committed to running agentic commerce on commercetools with Stripe as the payments layer.
The strategic positioning is clear: where Google aims to own the discovery-to-checkout flow, Stripe aims to own the payments infrastructure that every agent – regardless of surface – runs through.
Microsoft: Copilot Checkout and Brand Agents
Microsoft moved first, announcing its agentic commerce strategy on January 8 – three days before Google’s keynote. The centerpiece is Copilot Checkout, which lets consumers discover, compare, and purchase products entirely within the Copilot chat interface across Copilot.com, Bing, MSN, and Edge.
Microsoft shared performance data that caught attention: Copilot shopping journeys are 33% shorter than traditional search paths, 53% more purchases occur within 30 minutes when Copilot is involved, and sessions with shopping intent are 194% more likely to result in a purchase.
The second pillar is Brand Agents – AI-powered shopping assistants deployed on a retailer’s own website, trained on the brand’s product catalog, and operating in the brand’s voice. Currently Shopify-only and deployable in hours, Brand Agents handle product Q&A, upselling, cross-selling, and post-purchase recommendations. Alexander Del Rossa, a sleepwear retailer, reported 3x higher conversion rates in Brand Agent-assisted sessions versus unassisted ones. Microsoft’s own research found that consumers trust on-site brand agents 3x more than third-party agents, which frames Brand Agents as a complement to (rather than a replacement for) Copilot Checkout.
On the enterprise side, Microsoft announced four retail-specific agent templates in Dynamics 365 via Copilot Studio: a Personalized Shopping Agent, a Catalog Enrichment Agent, a Store Operations Agent, and the Brand Agent. The Dynamics 365 integration gives Microsoft a unique position – no competitor offers an equivalent end-to-end stack from consumer search to ERP.
OpenAI and Target: Shopping Inside ChatGPT
OpenAI and Target announced a dedicated Target shopping experience within ChatGPT. Customers can receive personalized recommendations, build carts, and complete checkout through the ChatGPT interface – a notable move for a top-10 U.S. retailer to make a conversational AI platform a first-class sales channel.
Separately, at its Dev Day, OpenAI outlined a broader framework for agents supporting end-to-end shopping: searching across retailers, comparing products, interpreting nuanced preferences, and completing purchases with participating merchants. Combined with OpenAI’s co-maintenance of ACP alongside Stripe, the company is positioning ChatGPT as both a shopping surface and a commerce infrastructure provider.
Shopify: Agentic Storefronts and the 11x Growth Signal
Shopify played a central role across multiple announcements. The company co-developed Google’s UCP, serves as the merchant platform for Microsoft’s Copilot Checkout and Brand Agents, and is an endorser of Stripe/OpenAI’s ACP.
Shopify’s own NRF messaging, themed “Commerce Favors the Bold,” emphasized that merchants on Shopify are automatically positioned for agentic commerce across Google, Microsoft, and OpenAI surfaces without building separate integrations. The 11x increase in AI search orders on Shopify stores since January 2025 served as the headline proof point.
For Shopify’s merchant base – particularly small and mid-market brands – the implication is significant: the platform is handling protocol adoption on their behalf, making agentic readiness a feature of the platform rather than an engineering project.
SAP, Salesforce, and commercetools
Three enterprise commerce vendors made notable announcements:
SAP introduced the Order Reliability Agent, part of SAP Order Management Services (planned for Q2 2026). The agent proactively identifies and resolves potential order issues – inventory mismatches, fulfillment risks, delivery exceptions – before they affect customers. SAP’s messaging at the show was direct: “For retailers, agentic commerce is here.”
Salesforce unveiled Guided Shopping v2, Contextual Search, Agentic Merchandising, and an OpenAI Syndication capability for scaling merchant productivity. Salesforce also provided some of the most-cited data from the conference: AI agents handled 142% more tasks during the 2025 holiday season compared to the prior two months, and AI and agents drove approximately 20% of holiday retail sales (Salesforce Commerce Cloud data).
commercetools launched its AI Hub – a connective foundation for agentic workflows – and was the first enterprise adopter of Stripe’s Agentic Commerce Suite. JD Sports publicly committed to executing agentic commerce on commercetools.
Key Stats from NRF 2026
The data points that anchored the conference narrative:
| Metric | Figure | Source |
|---|---|---|
| Share of 2025 holiday sales driven by AI agents | ~20% | Salesforce |
| Retailers planning agentic AI deployment in 12-24 months | 68% | NRF/industry surveys |
| NRF attendees implementing or planning agentic commerce | ~75% | Conference polling |
| Increase in AI agent task handling (2025 holiday vs. prior period) | 142% | Salesforce |
| U.S. consumers using generative AI for product research | 30-45% | Bain & Company |
| Projected U.S. B2C revenue orchestrated by agentic commerce by 2030 | $1 trillion | McKinsey |
| Projected global agentic commerce revenue by 2030 | $3-5 trillion | McKinsey |
| Copilot shopping journeys vs. traditional search | 33% shorter | Microsoft |
| AI search orders on Shopify stores (YoY growth) | 11x | Google/Shopify |
Retailer Concerns: Disintermediation, Data Readiness, and Governance
Behind the keynote optimism, retailers voiced substantive concerns in panels and private conversations:
Brand disintermediation. If consumers shop through AI agents rather than visiting brand websites, retailers risk being reduced to fulfillment suppliers. Maintaining brand identity, customer relationships, and margin control when an AI mediates the purchase decision is an unsolved challenge. This concern was a recurring theme in sessions like “Beyond the ‘Buy’ Button: Thriving in the New Agentic Commerce Era.”
Data readiness. Panelists consistently identified machine-readable product data as the single biggest hurdle for legacy retailers. Agentic commerce protocols – UCP, ACP, and the agents that consume them – require structured, enriched product catalogs with attributes that go far beyond traditional e-commerce feeds: compatibility data, substitute products, answers to common questions, and real-time inventory signals. Retailers without clean, comprehensive data will be invisible to agents.
Governance and security. Jack Hilger, Senior Director of Product at Visa, noted during a panel that current shopping experiences still resemble traditional e-commerce rather than truly native agentic flows, suggesting the industry is in early innings. Yelena Reznikova of OpenAI emphasized structured product feeds as essential for capturing intent. Multiple sessions addressed the new attack surfaces created by agents interacting autonomously across systems – including the potential for malicious code injection into AI platforms.
UX homogenization. Several retailers worried that agent-mediated shopping, if built solely on large language models without rich semantic data, would flatten the shopping experience into something generic. Differentiation in an agentic world requires investing in data depth, not just data availability.
What This Means for Merchants: Five Action Items
For retail executives processing the volume of announcements, here is what demands immediate attention:
-
Audit your product data. Every major protocol (UCP, ACP) and every AI surface (Gemini, Copilot, ChatGPT) depends on structured, machine-readable product feeds. If your catalog lacks detailed attributes, compatibility information, and real-time inventory, your products will not surface in agentic results. This is the single highest-leverage investment.
-
Evaluate your platform’s protocol readiness. If you are on Shopify, you are already enrolled in Copilot Checkout and positioned for UCP. If you are on commercetools, Stripe’s Agentic Commerce Suite provides protocol coverage. If you are on a legacy platform, assess the gap.
-
Own the on-site agent experience. Microsoft’s data showing 3x higher trust in brand-owned agents vs. third-party agents is a signal. Deploying a Brand Agent or equivalent on your own properties lets you capture intent data, control the conversation, and build a proprietary learning loop.
-
Prepare for payment token standards. Stripe’s Shared Payment Tokens and Google’s Agent Payments Protocol (AP2) are establishing how agents will handle money. Understand the implications for your checkout flow, fraud models, and payment provider relationships.
-
Develop an agentic governance framework. Establish policies for how your brand appears across AI surfaces, what agents can and cannot do on your behalf, what data you share with platforms, and how you measure attribution. The retailers who build this muscle now will have a structural advantage as agentic commerce scales.
Frequently Asked Questions
What is agentic commerce? Agentic commerce refers to AI agents that can autonomously discover, evaluate, compare, and purchase products on behalf of consumers. Unlike traditional e-commerce (where a human navigates a website) or conversational commerce (where a chatbot assists), agentic commerce involves AI systems that can execute multi-step purchasing workflows with minimal human intervention.
What is the difference between UCP and ACP? Google’s Universal Commerce Protocol (UCP) is an open-source standard covering the full commerce lifecycle – from product discovery through checkout and post-purchase order management. Stripe and OpenAI’s Agentic Commerce Protocol (ACP) focuses specifically on embedding commerce and payments into AI applications, with Shared Payment Tokens as its core innovation. The two protocols are designed to interoperate: Stripe’s Agentic Commerce Suite supports both UCP and ACP through a single integration.
Which retailers are already live with agentic checkout? As of February 2026, Etsy and Wayfair products are purchasable within Google’s AI Mode and the Gemini app. Urban Outfitters, Anthropologie, and Etsy merchants are live on Microsoft Copilot Checkout. Target has announced a dedicated shopping experience in ChatGPT. Shopify merchants are automatically enrolled in Copilot Checkout.
Do merchants lose control of the customer relationship? Both Google and Microsoft have emphasized that the merchant remains the merchant of record in agentic transactions. The retailer controls pricing, fulfillment, and customer service. However, the concern about losing influence over the discovery and decision-making process is legitimate – brands that do not optimize their data for AI surfaces risk being deprioritized by agents.
What does a merchant need to do to participate? At minimum: maintain a complete, structured product feed in the relevant merchant centers (Google Merchant Center, Microsoft Merchant Center), ensure your payment processor supports agentic protocols (Stripe’s Agentic Commerce Suite is the most comprehensive option today), and verify that your e-commerce platform is protocol-ready. Shopify merchants have the lowest barrier to entry.
How big is the agentic commerce opportunity? McKinsey projects agentic commerce could orchestrate up to $1 trillion in U.S. B2C revenue and $3-5 trillion globally by 2030. Salesforce data shows AI agents already drove approximately 20% of 2025 holiday retail sales. Bain & Company reports that 30-45% of U.S. consumers are already using generative AI to research and compare products before purchasing.
Is agentic commerce limited to the U.S.? Current deployments – Copilot Checkout, UCP-powered checkout in Google, the Target/ChatGPT experience – are U.S.-only. Google has announced global expansion “in the coming months.” The Agent Payments Protocol (AP2) has PIX (Brazil) and UPI (India) on its roadmap, signaling that real-time payment methods in emerging markets are a near-term priority.
Sources: Google Blog, Stripe Blog, Microsoft Industry Blog, Salesforce Commerce Cloud, SAP Newsroom, commercetools Press, NRF 2026 session archives, McKinsey, Bain & Company. All announcements from NRF 2026 (January 8-12, 2026) and subsequent launches through February 2026.
Hexagon Team
Published March 8, 2026
