The 3 AI Terms That NRF’s Big Show Suggests Are Going To Be Big in 2025
Even if you’ve been living under the proverbial rock for the last few years, there’s a pretty good chance you’re using ChatGPT to summarize and prioritize your email inbox. Nobody will have been surprised that artificial intelligence (AI) dominated the agenda for the National Retail Federation’s “Big Show” two years running, in both 2024 and 2025.
This year felt different though. Where 2024’s show was full of optimism and invitations to jump in and ‘get going’ with AI, there was an open question left hanging in many of the panels and conversations when it came time to discuss specific use cases. In 2025, something like 30 speaking sessions invoked AI in the title. What changed this year was the tangible feeling that retail has ‘got going’ when it comes to AI deployments.
We heard three phrases repeated across the three days of the conference. They’re going to be important for retailers this year in their AI transformations, large or small.
- Change management
- Data quality/governance/unification
- Agentic AI
Of course these terms on their own aren’t new, with the possible exception of “Agentic”, which Google Trends data suggests exploded in popularity around October 2024.
The point is that these AI concepts’ predominance and importance at a conference like this shows that retail leaders have largely shifted from a broad enthusiasm for AI to active engagement with its implementation. Here’s where these three terms cropped up and what they mean for retail in 2025.
For a detailed discussion of AI adoption in retail supply chains, you can watch and read the transcript from Blue Yonder’s Big Ideas session from the conference, and for insights on how supply chain organizations are transforming in response to the AI revolution, read our Spotlight Paper.
Change management
This could have been a talk track of its own at the Big Show and was brought up by retailers across verticals, from grocery (Sprout Farmers Market) to apparel (Tailored Brands, DXL) to jewelry (Pandora) to beauty (Ulta Beauty) and more. It was perhaps the most consistently asked question in panels, even those whose topic was not primarily AI.
Many acknowledged the fear that AI will automate jobs away, and these speakers were near universal in their desire to reassure the audience that AI adoption does not mean reducing headcount. Instead they focused on reducing toil, laborious and/or repetitive work, allowing associates to work more productively and decisively on higher-level tasks.
That was exemplified in a comment from Rob Bogan, Chief Technology Officer at DXL Group, who joined Blue Yonder’s Chief Innovation Officer Andrea Morgan-Vandome in conversation with Nik Haggerty, Head of Retail, Supply Chain & ERP Technology at Chalhoub Group for our Big Ideas Session.
“We want merchants to be merchants … back in the day they would spend a month as data analysts getting ready for line review, and then in the line review they couldn’t answer a strategic question because they’re so mired in the data”
So, retailers like DXL Group are handling the adoption and transformative power of AI through targeting it at the kind of work that tends to obstruct strategic and productive thinking. Others spoke about the importance of earning trust, particularly in a use case where AI is making or suggesting decisions.
Chalhoub’s Nik Haggerty mentioned that people find it hard to trust ‘black box’ models, where it’s hard or impossible to see why the decision has been made a certain way.
“The biggest challenge is about trusting the answer [given by AI] … the answer just appears and they can’t actually work out how. It becomes a change management conundrum.”
Building trust within an organization in the decisions and suggestions of AI is critical. Without this, AI implementation will fail to drive productivity improvements as users obsess about how decisions are reached rather than how best to utilize them. Nik went on to discuss the principle of explainability as a means of alleviating the trust issue:
“I think one of the things that vendors should be helping with is the concept of ‘clear-box’ AI, where you can see inside because as soon as you can explain the magic underneath, that helps people to embrace and understand it.”
It’s clear that AI is now starting to be embedded in retail organizations, and the focus is on effective implementation, cultural acceptance of AI, and the right principles and approaches to help smooth out the transition.
Data: unified, governed, and high-quality
“AI is only as good as the data over which it reasons.”
That was Microsoft’s Shelley Bransten, CVP Global Industry Solutions, in her talk on retail-ready AI, and it sums up the extent to which data quality, accessibility and governance were practically ever-present topics at the Big Show.
If AI is to transform retail, data will be the limiting factor for how much the industry can evolve. Data suggests that just 29% of business applications are integrated. In such an environment, there’s a lot of work to do before AI can realize its transformative potential.
Baltazar Hasselsteen Ozonek, VP AI & Innovation at jewelry retailer Pandora, offered an interesting perspective on the perennial nature of ‘data quality’ as a topic in tech transformations.
“We haven’t really cracked the data quality thing yet. We didn’t do that when we launched dashboards a decade ago, and we didn’t do that when we launched spreadsheets three decades ago. And so we still need to fix the foundations.”
However, while data accessibility and quality are limiting factors for AI adoption, some AI technologies are helping to make improving data and accessing data much easier. Large Language Models (LLMs) were frequently mentioned as having the potential to ingest data in new ways that strictly machine learning tools aren’t typically able to, for example being able to use unstructured data like free text, as well as the potential for LLMs to support data preparation leading to improved data quality. That capability could be key in helping retailers get their data ducks in a row.
The importance of data not just being high-quality but truly unified was also brought up by retailers around the show, and featured particularly in Blue Yonder’s session: “The supply chain revolution: Can your organization succeed in the age of AI?”
Both retailers spoke about the value of Blue Yonder’s unified data platform in enabling them to quickly integrate data from other applications into their supply chain planning and merchandising suite, and the flexibility and speed such platforms offer.
Nik Haggerty cited the example of upgrading other applications, in the case of Chalhoub Group, their Enterprise Resource Planning (ERP) system. Thanks to Blue Yonder’s unified data approach, Chalhoub was able to move ahead with implementing Blue Yonder’s end-to-end platform without needing to wait for the new ERP to come online.
Agentic AI
AI agents have quickly become a very popular topic. They’re rapidly being introduced and integrated into existing software applications, and offer a very powerful new way for users to interface with tech. Typically AI agents enable automation of tasks and workflows in ways that are more flexible and adaptable than rules-based automations and workflow designs have historically been able to achieve.
That’s partly because agentic AI draws on new technological elements like generative AI to both offer a natural language interface to users, as well as to interpret and analyze documents, data and systems. That means AI agents are capable of understanding text-based instructions, developing and executing a plan to carry out those instructions, and applying that plan to data and systems on the user’s behalf.
All of which is to say that ‘agentic’ AI appears to be the vector by which a lot of the exciting technical innovations of the last several years will actually manifest for many retailers. DXL’s Rob Bogan discussed the transformative potential of AI agents:
“The magic is, a merchant or a planner can actually ask questions of an agent. ‘What are my 10 best attributes that I should consider for the next fall season’, or ‘which extra stores should we invest in’, or ‘have I got a problem in men’s shirts?’. And a human could do that, but it would take them two or three days… And so an agent wouldn’t replace people, it augments them.”
The ability of AI agents to understand strategic questions like these, quickly analyze multiple data sources, to make recommendations in response to them, and then to actually action those recommendations across multiple systems, is a game changer. Not just for productivity, as retailers around the show pointed out – this means strategic questions get asked, investigated, modelled and assessed far faster, allowing for much greater flexibility and a serious competitive advantage.
Is your retail organization adapting to adopt AI? Find out how retail leaders are planning their organizational changes in response to the AI technology revolution, including the role of agents, in our Spotlight Paper, Re-Organizing for AI: How Supply Chain Leaders Have to Adapt.