Retail is detail. And merchandising sits at its heart because the decisions it governs shape almost everything a retailer does—what to sell, how much to pay for it, how to price and promote it, how much inventory to hold, where to allocate it, and how to collaborate with suppliers.

Bain’s recent research on the future of retail highlights how operational complexity is challenging the ways core functions operate. Despite expanded product assortments, proliferating channels, increased shopper expectations, and technology disruption, many merchandising teams still rely heavily on spreadsheets and manual effort. The problem is not data access but the capacity to fully understand it and make consistent, high-quality decisions.

The stakes are rising further as generative AI tools such as ChatGPT promise to disrupt both strategic processes and day-to-day business management. Just as autonomous “AI shoppers” are beginning to research, compare, and buy on behalf of customers, AI agents can transform how retailers keep pace.

To take advantage of this, leading companies are moving beyond experimentation—using AI not just to understand the landscape but to make, and increasingly automate, decisions across the business.

What AI really changes

Generative AI tools can already perform work that once consumed endless hours for category teams, such as scanning performance reports, summarizing shopper reviews, and tracking competitors. Retailers are beginning to reap these benefits. French retailer Carrefour has been applying generative AI to procurement workflows to speed up tasks such as supplier quote comparisons. This frees up time for higher-value work and delivers tangible productivity benefits. Broadly, Bain & Company projects have shown that applying AI solutions can create efficiency gains of up to 50% to 70% for key tasks while also increasing merchants’ AI literacy.

Beyond this, agentic AI opens up incredible potential for retailers. Whereas traditional analytics tools could describe what had happened, they couldn’t reason across fragmented data, adapt to changing context, or support decisions as work unfolded. Agentic AI changes this. These tools combine generative reasoning with orchestration models that can monitor data, reason across signals, and recommend (or take) actions autonomously within guardrails set by the business to ensure appropriate human oversight.

Agents can process large volumes of structured and unstructured data—ranging from competitor catalogs and product attributes to customer sentiment and search behavior—to surface emerging risks or opportunities. Over time, they evolve from summarizing data to recommending actions and, eventually, executing many of these autonomously.

This shift is about more than speed. It represents a move from spreadsheets and dashboards to collaborative decision agents embedded at the heart of retail.

Applying agents in merchandising

Agents can reshape both strategic category management and everyday performance. At the strategic end, AI helps merchants rapidly build customer-centric category strategies by synthesizing performance data, shopper insights, and external signals. By simulating thousands of customer behavior scenarios, agents can identify duplication, gaps, and unmet needs. Similar transformations apply to pricing, promotion, and supplier strategy. By consolidating supplier information in one place, AI enables merchants to prepare for negotiations, compare commercial terms, and draft agreements in as little as one hour, compared with days previously. One European grocery retailer, for example, has improved own-brand development by using AI to simplify specifications, compare suppliers, and manage cost pressure amid inflation.

Closer to execution, agents can support the adjustments needed as plans meet reality. For example, an agent might spot that sales in a particular category are below expectations, check whether competitors have cut prices or customer sentiment has shifted, and then recommend potential next steps. Or an agent could help decide when and where to mark down products, improve inventory allocation by flagging stock imbalances, and spot emerging micro-trends across social, review, and competitor data.

US-based multinational Walmart, for example, has embedded agents into daily routines. Its Wally assistant helps interpret performance data and answers operational questions in seconds, bringing AI directly into merchandising decisions.

Translating AI into action: scaling value

Retailers seeing the most success with AI start with practical, high-impact use cases, build the right data and architecture, and redesign workflows and roles so AI is used consistently and effectively. Following a “crawl-walk-run” model can deliver tangible value within three to six months, using existing data and workflows rather than waiting for major system overhauls.

The simplest entry point is often equipping merchants and planners with general-purpose tools such as ChatGPT to build AI literacy and make existing processes more efficient, from data analysis to report writing and competitive intelligence.

In the crawl phase, AI is connected to data for a specific decision, such as refining a category strategy using recent performance, shopper insights, and competitor context. Tractor Supply, the rural lifestyle retailer, shows how this can scale. The firm deployed ChatGPT Enterprise, supported by AI champions across multiple departments, and generated hundreds of use cases, from quickly turning loss-prevention information into actionable insights to reducing time spent accessing and analyzing other data.

In the walk phase, AI is connected to more data sources and steps in the workflow, enabling agents to recommend assortment, promotion, or pricing actions. Target, for example, has worked with Bain and OpenAI to build AI agents that help store teams with operations and customer interactions, as well as within merchandising to better understand category performance ahead of vendor discussions.

In the run phase, agents are wired directly into execution systems, enabling AI-initiated pricing, promotion, or inventory changes with human oversight. Bain’s AI Retail Merchant Assistant, for example, can embed AI into merchandising workflows. It can consolidate performance, trend, and supplier data to support strategic choices on assortment, pricing, and promotions, as well as surfacing role-specific insights that accelerate confident decision making across the merchandising life cycle.

Building strong foundations

Navigating the shift to the run phase—agentic AI—requires the right foundations. Retailers must connect core internal data, such as point of sale, margin, inventory, loyalty, and supply chain execution, with relevant external signals including competitor activity, digital behavior, and sentiment. 

A generative AI reasoning layer, supported by an agentic orchestration layer that calls forecasting, optimization, pricing, and workflow tools, sits on top of this.

Technology alone, however, is not enough. While generative AI can be applied effectively to existing systems, adding agents to inefficient processes only exacerbates challenges. Successful organizations rethink how merchandising work gets done—from the role of the merchant to the end-to-end merchandising process—as part of their efforts to embed AI.

This means reshaping roles and routines. As AI agents take on more of the analytical heavy lifting, merchants can spend less time assembling data and more time on judgment and strategy. New roles will emerge, such as AI product owners responsible for shaping use cases, prioritizing features, and ensuring tools fit merchant workflows. Core processes, including category reviews, sign-offs, and vendor meetings, will also increasingly change, starting with agent-generated scenarios rather than static presentations.

Accelerating momentum to scale

A practical checklist for building momentum begins with defining an agentic AI and decision matrix, describing which decisions can be automated, which require review, and which remain human-only.

Choosing two or three pilot use cases linked to quantifiable outcomes, such as promotion optimization or assortment localization, and carrying out a data inventory are next steps.

Senior executive sponsorship matters if changes are to stick. The chief merchandising officer or equivalent must own the merchant AI roadmap, with change management and training embedded in live work rather than treated as a separate program. Over time, AI literacy should become a core merchandising capability, something expected of every merchant, rather than a specialist skill held by a few experts.

Defining key performance indicators, reporting cadence, and pilot review timelines also pushes progress forward. Finally, governance is critical: What guardrails need to be in place, what circumstances would trigger a rollback of any AI-related pilots or processes, how will AI use be audited, and how is this all laid out in vendor and supplier policies?

The opportunity—and the risk of waiting

Retail is already in its AI era. Early deployments are delivering measurable gains, but the greatest value will accrue to retailers that translate these outcomes into a fundamentally new merchandising model.

Merchandising is a natural starting point for broader transformation because decisions made here cascade into supply chain, store operations, labor, and digital execution. Redesigning merchandising processes is therefore often the fastest way to drive AI improvement across the rest of the operating model.

What comes next extends beyond merchandising into store operations, shaping workforce tasking, deployment, and on-shelf execution. The path forward is clear: Start with practical use cases tightly linked to profit-and-loss impact, build the data and orchestration backbone, and redesign merchant roles and processes so AI is embedded into how decisions get made. Retailers that move deliberately and at scale will convert early productivity gains into decisively better category strategies, stronger outcomes, and new commercial opportunities. Those that wait risk being shaped by others’ agents rather than owning their future.

OpenAI

OpenAI is an AI research and deployment company. Our mission is to ensure that artificial general intelligence benefits all of humanity. Founded in 2015, OpenAI develops cutting-edge AI technologies, including the GPT series of language models, and partners with organizations to integrate AI capabilities into real-world applications responsibly. OpenAI is committed to building safe, ethical AI systems and fostering transparency, safety, and alignment across the global AI ecosystem.