Introduction

Coresight Research is a research partner of Shoptalk Spring 2026, which took place March 24–26 at the Mandalay Bay resort in Las Vegas, Nevada. In this report, we present our wrap-up of the conference, drawing on the many sessions that our team attended and discussions that our team participated in across the three days.

Under the overarching theme “Retail in the Age of AI,” the event explored how AI is reshaping every element of the retail ecosystem—from back-office operations and supply chains to personalization, marketing, physical stores and customer engagement. At the same time, the agenda explored the roles of emotion, experience and human connection as counterbalances to automation.

Shoptalk Spring 2026 Wrap-Up: Coresight Research Analysis

1. Agentic Commerce and AI: A Broad Landscape of Views, Predictions and Uncertainties

The Promise of AI in Retail

• Bret Taylor (Co-Founder & CEO, Sierra, and Chairman of OpenAI) posited that AI agents will become as essential as a retailer’s website or app, acting as the “digital front door” to brands. In his view, AI is serving as an extremely valuable tool for customers who want swift, efficient service (like checking order status or tracking deliveries) without needing human intervention. Taylor argued that applying AI agents in functional areas can free up more of a retailer’s “budget for service” to spend elsewhere on high-touch experience.
• Matt Nichols (General Partner, Commerce Ventures) predicted that agentic commerce will drive e-commerce growth, particularly through the automation of complex decision-making processes. However, he pointed out a missing piece in agentic commerce: the lack of a universal API for accepting agentic orders, something that needs to be resolved for wide-scale adoption.
• “Shopping is moving from product search to problem solving, and that changes both the interface and the economics of discovery,” according to Mahak Sharma, Product Partnerships, Search & Commerce, OpenAI. Sharma described how shopping behavior inside ChatGPT is evolving: consumers increasingly begin with goals, constraints and context rather than specific products, expecting curated, personalized responses instead of static lists. The interface must dynamically adapt to different missions, from high-consideration purchases to event-based planning or visual discovery. The launch of new shopping capabilities within ChatGPT—including comparison-led experiences, image-based search and flexible merchant pathways spanning on- and off-platform checkout—demonstrates how commerce journeys are becoming model-native and fluid, rather than fixed and linear.
• In agentic commerce, “the risk in not investing is greater than investing and getting it wrong,” even though retailers do not yet know exactly what will happen with agentic AI, according to Ben Miller VP, Original Content & Strategy, Shoptalk.

The Reality of AI in Customer Service

• Angie Brown (EVP & Chief Information Officer, The Home Depot) emphasized how AI assistants can enhance customer experiences by helping store associates with real-time product compatibility and troubleshooting, enabling them to spend more time with customers. However, she pointed out that AI’s role in customer service should be supportive, not a replacement for human interactions: “Agentic will be an “and” to an ecosystem of solutions.”
• We heard similar from Macy’s, which is “not in the business of transacting, but for moments that matter,” according to Max Magni, Chief Customer & Digital Officer, Macy’s Inc. Those moments can begin online and finish in-store and Macy’s is deploying AI as a part of that cross-channel experience, with its Ask Macy’s chatbot and with greater personalization. Yet the differentiator for Macy’s is its people in stores, and AI tools are enabling store staff, including store managers, to spend less time on repetitive tasks such as reporting and more time with customers on the shop floor.
• Similarly, at The Vitamin Shoppe, a “Shop Advisor” surfaces relevant educational content and product recommendations in-store, empowering both customers and associates to navigate complex health and wellness decisions.
• Sven Gerjets (EVP & Chief Technology Officer, Gap Inc.) outlined Gap Inc.’s approach to AI as one grounded in structure rather than hype, with the company’s “Office of AI” organized around the customer journey and focused on identifying where work should be reimagined, not simply automated. Gap is using AI to rethink workflows, data foundations and decision-making at scale, particularly in areas with clear value potential such as product discovery, checkout and sizing. Gerjets emphasized that companies cannot afford a wait-and-see posture. That urgency helps explain Gap’s early partnership around Google’s UCP protocol, which is designed to let merchants bring their own experiences into LLM-driven commerce journeys. For Gap, the strategic value is not just visibility inside AI interfaces, but preserving merchant control over relevance, customer data and the end-to-end brand experience.

Andrew Laudato, Chief Operating Officer, The Vitamin Shoppe
Source: Shoptalk Spring

 

Some Skepticism About Full Agentic Commerce Adoption

• Janie Yu (Partner, LFX Venture Partners) stressed that while agentic commerce has tremendous potential, but monetization models remain unsettled—for example, whether platforms will take a share of sales (GMV) or adopt an advertising-focused model. The latter could be akin to current search-monetization models—closer to Google Search than Amazon.com.
• Joe Laszlo (Head of Content & Insights, Shoptalk) argued that agentic commerce (or “a-commerce” as Laszlo termed it) might be overhyped, particularly the belief that bots will do all shopping for consumers. In his view, while agentic AI will improve behind-the-scenes processes, it will not replace human interaction in the shopping experience, especially for lower-consideration purchases.
• On personal agents: “The more considered the purchase, the more consumers will use personal AI to help them with the decision,” said OpenAI’s Bret Taylor. “Not every retail experience will be driven by an individual AI agent.”

The Competitive Landscape

• Matt Nichols predicts AI’s role in retail will not just lead to efficiency but could also reverse retail concentration by helping smaller brands scale faster and compete with retail giants such as Amazon, which may be overwhelmed by a new wave of nimble, AI-powered startups.
• Ben Miller highlighted that AI agents would contribute to e-commerce growth, and also predicted that “a-commerce” will empower new brands to come to market faster and be more nimble, which will fragment the retail market.

Future Outlook: Personalization and Data

• Jill Smith (CEO, IRIS) mentioned how AI can create deeply personalized experiences by using real-time data to understand customer needs, aligning with Bruce Richards (Head of Industry Strategy, Adobe), who also sees AI-driven personalization as the next frontier for improving customer journeys.
• However, Janie Yu argued that for agentic commerce to become ubiquitous, companies must resolve the challenge of data readiness, ensuring that AI systems can handle large amounts of unified data to provide truly personalized experiences.

2. AI Technology Recently Passed a Key Inflection Point

In an AWS-sponsored breakfast, Ryan Ball of Anthropic observed that the nature of AI has changed dramatically over the past six months, due to improvements in model capabilities that turned them from question-and-answer assistants to agents that can do deep research work on humans’ behalf. Further, AI deployment requires a two-pronged approach comprising top-down clarity while enabling bottoms-up adoption. In conversations about AI today, model capabilities are not the limiting factor—rather, the limitations follow from the data input to the model.

3. The Power of Agentic AI Is Forcing New Build-Versus-Buy Decisions

The advent of new AI-based technologies, such as agents for writing computer code, is resulting in companies receiving phone calls from former technology companies and consultants offering to create solutions to replicate existing SaaS (software as a service) platforms in as few as two weeks. These offers fail to take into account that an entire company spent years or more in developing these solutions. Robert Ibarguen, Head of Digital Intelligence at Newell Brands, commented that users could use an LLM to replicate the functionality of a platform such as that of Salsify—but badly. While an IT team could build a solution, can they also maintain it, grow it and scale it? These steps require people and domain expertise.

The deployment of AI rests on having high-quality data, yet data gaps are inevitable. One solution is the use of synthetic data to bridge data gaps, yet the superior solution is to build a system with good data hygiene so that the enterprise only has to clean up its data only one time.

4. AI Can Help with Intimidating Home-Improvement and Furnishing Decisions

Taryn Dominie, Head of Industry, Orange Apron Media, The Home Depot, commented that the DIY (do it yourself) journey is complex and that consumers get stuck in “color paralysis,” i.e., being unable to select a color. To combat this, the retailer has launched a new AI tool called Chat Hue to make the color-selection process as seamless and interconnected as possible. In addition, The Home Depot’s Magic Apron AI tool helps the consumer throughout the shopping journey and seeks to offer an “Orange Apron” experience on site through being a color authority.

Similarly, Niraj Shah, CEO & Co-Chairman of Wayfair, commented that, although home is a “fun” category for customers, some find it overwhelming, and AI can help to guide the customer easily and intuitively. Some of the least-convenient areas of a furniture purchase, are delivery, assembly and installation, and AI can ease the pain.

Although generative AI may now seem outmoded following the advent of agentic AI, there are still areas in which the technology can offer significant value for customers, specifically in populating product pages for retailers with large portfolios. Wayfair, for example, mentioned its product catalog has 20 million SKUs, which AI can populate automatically, in addition to preventing customer aggravation when product dimensions are incorrect. Newell Brands commented that the volume of constant PDP (product description page) changes was impossible to maintain manually. Its Bonsai AI content agent was customized within 80 days and achieved a 40x improvement in productivity.

Left to right: Jordan Broggi, EVP of Customer Experience and President of Online, Home Depot; Angie Brown, EVP & Chief Information Officer, Home Depot; and Vidhi Choudhary, Retail Reporter, Morning Brew
Source: Shoptalk Spring

 

5. Video Can Make Shopping a Communal Activity

Geoffrey Goldberg, Chief Creative Officer & Co-Founder, Movers+Shakers, equated earned attention with cultural relevance. Some creator choices have unexpectedly created cultural relevance—for example, the viral pimple patch, which connected cultural conversations with commerce, generating 31 million views, 25 million new TikTok followers and more than 3 million engagements.

Tom Verrilli, Chief Product Officer of Whatnot, announced that, in contrast to common opinion centering on a specific generation, consumers across all major age groups are enthusiastic live shoppers. Despite relatively few Western consumers experimenting with live shopping, Whatnot viewers spend from 20 minutes to more than one and a half hours daily on the platform, averaging 95 minutes a day, which he claimed exceeded Netflix viewing. He views shopping as a social activity, like going to the mall with friends, which involves community, a transaction and building a brand—all simultaneously. Whatnot also builds authenticity and trust, as purchasers can see the faces of sellers, in contrast to e-commerce. Although the company started in the collectible space, today consumers can buy bags, sneakers, sports cars, fresh food—even live lobsters—replicating the mall experience.

6. Technology, Including Drones, Continues to Improve the Delivery Experience

Technology can help drive down the cost of failed deliveries. Around 8% of first-time deliveries fail, with each late or failed delivery costing between $17 and $40 once reattempts, support tickets and customer churn are factored in, according to Salman Habib, Co-Founder & CEO at Burq. However, AI agents for automated exception recovery can bring efficiency in areas such as auto-rerouting. Habib made the case that while every company tracks cost per delivery, they should start tracking cost per failed delivery, not least because the customer who blames the brand and doesn’t return is the hardest cost to measure and the most damaging.

Meanwhile, drones are adding to the quick-commerce ecosystem. Heather Rivera, Chief Business Officer at Wing, said Wing grew five times in the last year, driven by a shift toward small-basket orders wanted quickly. With flight times under five minutes, Wing’s partners are already seeing gains in purchase frequency and customer satisfaction. Rivera was clear that drone delivery is for everyone but not for everything, and it won’t replace the heavy Sunday grocery shop, but it is well-suited to a missing ingredient for dinner or an over-the-counter medication for an aging parent.

In the “Retail Zeitgeist,” Shoptalk’s Laszlo and Miller predicted that US retail would see a new cycle of investment in fulfillment. The very strong growth in grocery e-commerce and the shift toward quick commerce will make investments, such as in micro-fulfillment centers, worthwhile. Our research on grocery e-commerce and fulfillment trends supports such a conclusion.

Left to right: Joe Laszlo, Head of Content & Insights, Shoptalk; Sarah Engel, President, January Digital; Ben Miller, VP, Original Content & Strategy, Shoptalk; and Chris Walton, President & CEO Omni Talk
Source: Shoptalk Spring

 

What We Think

The Coresight Research View on AI

We are entering a new phase of AI development driven by advances in agentic AI and physical AI. The first generation of AI—machine learning (ML)—centers on the ability of AI to find relationships, including hidden ones, in large datasets, and excels at prediction and optimization. The steady improvement in computing power, driven by Moore’s Law, and the resulting decrease in the cost of computing, aided by cloud computing, have enabled greater AI/ML capabilities. Today, AI/ML still delivers productivity gains and optimization benefits for retailers and other enterprises.

The advent of GenAI, the second generation of AI, adds the ability to communicate in natural human language and handle unstructured data, such as text, audio and video. Early GenAI applications focused on text generation and summarization, as well as natural-language chatbots. Multimodal models now enable the analysis and creation of voice and video, as well as synthetic data and computer code. We have also seen the creation of applications that manage multiple language models and the formation of businesses built entirely on GenAI technology. Combining ML for prediction with GenAI for its ability to generate multiple media outputs offers many powerful new capabilities.

Agentic AI builds upon GenAI’s ability to understand natural human language, which is used to create the instructions that agents follow. Agents operate by activating other software, following instructions and acting proactively on behalf of humans. We are still in the early stages of the agentic AI revolution, with major cloud computing platforms offering developer tools and a few enterprises just beginning to roll out agents. The future promises AI agents everywhere, and their value could increase even further with agent-to-agent communication. Most recently, AI chatbot providers have enabled retailers to share product information so that it is searchable by language models, and they now allow agents to help shoppers complete purchases from within the chatbot.

Agentic commerce could dramatically transform retail. Shopping agents do not represent an incremental change: they collapse the sales funnel by letting users make decisions within a chat session with the AI agent.

Physical AI and world models are expected to emerge as the next wave of AI development in 2026. Robotics with edge AI and physical AI, equipped with knowledge of the real world and motion, will transform supply chain and retail operations.

Implications from This Report

The path for agentic commerce is still being determined and that was reflected in discussions at Shoptalk Spring. Insights from the conference that are worth restating include:

• Adoption for factual information-finding is here, is proven and this is liberating some of retailers’ “service budget” for higher-touch interactions.
• Future consumer demand for purchasing through agentic apps and autonomous shopping agents remains unknown, and monetization of platforms such as ChatGPT could primarily be via advertising, like search and social media before them.
• Personal AI agents will probably be used most for complex, research-heavy purchases.

Brands or Retailers Poised to Gain Advantage

• Macy’s, The Home Depot and The Vitamin Shoppe are using AI tools to empower both employees and associates. Gap Inc.’s “Office of AI” is organized around the shopper journey and focused on identifying where work should be reimagined.
• Newell Brands is using generative AI to keep its extensive product catalog complete and up to date.
• Wayfair and The Home Depot are using generative AI tools to mitigate pain points from the home furnishing and decorating process.

Brands or Retailers That Risk Losing Advantage

• Retailers and brands that do not find their unique needs and applications for AI technology could find themselves at a competitive disadvantage.
• Neglecting the capabilities of agentic AI and agentic commerce could put retailers at a further disadvantage.

Notes

Data in this report are as of April 8, 2026

Companies mentioned in this report are: Amazon.com, Inc. (NasdaqGS: AMZN), Anthropic, Burq, Commerce Ventures, IRIS, LFX Venture Partners, Macy’s, Inc. (NYSE: M), Movers+Shakers, Newell Brands Inc. (NasdaqGS: NWL), OpenAI, Shoptalk, Sierra, The Gap Inc. (NYSE: GAP), The Home Depot, Inc. (NYSE: HD), The Vitamin Shoppe, LLC, TikTok, Wayfair Inc. (NYSE: W), Whatnot Inc., Wing.