{{item.videoDuration}}
{{item.title}}
{{item.text}}
{{item.videoDuration}}
{{item.text}}
There was a time when the leading merchants in retail were the industry's tastemakers. They didn't just buy products. They defined culture. They read the room before the room knew what it wanted, placed bets on color, fit, and category, and turned those bets into the trends that defined a season. A great merchant wasn't following the market. They were leading it.
Then something happened. Retail evolved. Channels multiplied. Data exploded. SKU counts ballooned into the hundreds of thousands. And the customers changed, too. Gen Z and Gen Alpha, digitally native generations who redefined how people research, engage, and transact, didn't shop the way prior generations did. They moved fluidly across platforms, expected personalization at scale, and made purchase decisions through signals that legacy systems weren't built to read. Somewhere in the process of managing all of it—the spreadsheets, the legacy systems that never quite talked to each other—the merchant got sidelined. The job that once required vision became a job that required stamina. Swivel-chairing between screens. Chasing approvals. The visionary became a data jockey who might spend their entire Monday piecing together last week’s report. That shift cost retail more than it realizes. But AI is about to give it back.
From isolated AI experiments to end-to-end value
Retail leaders are under real pressure right now. Boards and investors are taking the disruptive potential of AI seriously, and no executive wants to fall behind. But pressure can produce the wrong kinds of decisions—particularly in an environment where hype is loud and genuine signals of progress are harder to read. The retailers who are pulling ahead aren't the ones running the most experiments. They're the ones who have made the clearest strategic choices about where AI and agentic co-workers can create the most value. That calculus points to one function above all others: merchandising.
When AI is used for demand planning, promo optimization, item onboarding, and vendor funds management, retailers typically see a 3 to 5% revenue uplift and 15 to 30% in cost savings.
PwC analysis of AI benchmarking databaseThe autonomous merchandising operating model
Picture a fully autonomous merchandising operation. A series of specialized agents run the end-to-end flow, making data and insight-driven decisions on procurement, personalized and localized assortments, pricing and promotions, vendor item maintenance, purchase order reconciliation, and distribution. No spreadsheets. No manual exception handling. No Monday morning spent reconstructing last week's numbers. The merchant isn't buried in operational overhead. They're setting strategy, managing the agent team, and focusing their judgment on the creative and relational decisions that no algorithm can make.
That's the destination. Of course, getting there isn't a single leap. It's a deliberate progression along a spectrum, and knowing where to enter that spectrum, and how to move along it, is what separates retailers who capture compounding value from those who collect isolated wins.
The risk of getting this wrong is also real, in both directions. Hand too much authority to an agent without human creative input and the assortment narrows to what has already sold, greatly reducing the novelty that makes a brand worth shopping. The goal is to concentrate human judgment where it's genuinely irreplaceable, and to be rigorous about where it isn't.
Why merchandising first
Merchandising sits at the center of everything a retailer does. It is simultaneously the art and the science of the business—the function that decides what gets made or bought, in what quantities, at what price, placed in which channels, for which customers. Get it right and you compound gains across the entire P&L. Get it wrong and no amount of marketing spend or supply chain efficiency can safeguard your margins.
Consider what a merchant actually faces on a Monday morning. The weekend is over and the business has moved. What happened at that store in Iowa? Why is margin down? Why are two stores across the street from each other performing so differently? Why isn't the search algorithm surfacing the right product? Why is this color selling out in the Southeast but underperforming in the Northeast? Each question lives in a different system. And today's customer—shopping across social media, flagship stores, and third-party marketplaces, expecting the right product at the right price, available now—has made each one of those questions harder to answer.
Take something as routine as vendor shipment management. When a retailer receives an unexpected overship, that exception routes through a buyer who manually checks open-to-buy, weighs the inventory implications, and coordinates across warehouse and vendor teams. That overall workflow—inbox to resolution—can now be automated end-to-end. What used to consume hours of a merchant's week becomes invisible overhead. This is what it looks like with AI as a workflow that runs reliably at scale.
The return of the merchant
Here's the counterintuitive thing about applying AI agents in merchandising: it doesn't diminish the merchant. It restores them.
Those concerns about handing over the craft are legitimate. Merchants have spent careers developing judgment that no algorithm can replicate, and skepticism about whether AI outputs can be trusted persists. To address those concerns directly, start with the workflows where data quality is strongest and the stakes for errors are lowest.
When AI takes on the data work, what's left is the restoration of merchant judgment and vision. The merchant of the future manages a team of agents—pricing, allocation, trend research, vendor communications—each handling a discrete domain. They set the strategy, review the exceptions, and focus judgment where it matters most: the customer, the product, and the creative bets that no algorithm can make. Merchants freed from operational overhead are more powerful than they've been in decades.
The retailers that will gain competitive ground aren't layering AI onto existing ways of working—they're redesigning those functions from the ground up, starting with the question: if AI could handle this entirely, what would we build?
The leapfrog moment
Retailers who have simply layered AI onto existing infrastructure are finding that isolated wins don't compound. Transformation isn't additive. It's systemic. It requires deliberate integration to connect data, workflows, controls, and operating model choices so that AI is deployed coherently, not scattered.
The retailers who will likely gain competitive ground in the next three to five years are asking different questions: not “Where can we add AI?” But rather, “What would we design if AI could handle this function entirely?” Some are already acting on those questions, quietly shelving long-planned legacy system implementations in favor of AI-native approaches—and for good reason. Merchants don't want another multi-year digital transformation. They want to see change fast and to realize the value quickly.
The window to move ahead of competitors is open. But it won't stay open indefinitely. Newer entrants without legacy infrastructure are already operating with leaner, faster, more AI-integrated models. The structural advantages that set established retailers apart for decades are no longer as significant.
Where to begin
The next question to ask? If AI could handle this entire function, what would we design? Then work backwards to identify where human judgment is truly irreplaceable and safeguard that space. In merchandising, the starting points are clear:
- Map the overall merchandising workflow and identify which decisions require cultural context, creative vision, or relationship judgement and which are fundamentally data problems.
- Automate item induction, product data enrichment, and vendor exception management.
- Build AI-driven inventory allocation and demand forecasting.
- Deploy real-time pricing and promotion optimization.
- Let the online channel run on continuous automated merchandising logic.
- Define the new operating model: agent thresholds, human review triggers, and audit protocols for when agents underperform or fail.
This is about enhancing headcount rather than reducing it. Free merchants from operational overhead and restore them to the work that only they can do. This is the end-to-end value chain opportunity. It starts where it always has—with the merchant, the product, and the conviction that getting those right changes everything that follows.
Build your industry edge
Explore the end-to-end AI opportunity in retail—and where to start
{{filterContent.facetedTitle}}
{{contentList.loadingText}}
{{contentList.loadingText}}
Follow us
