Everyone is talking about agentic buyers. What about merchandisers?
Agentic AI is making waves in commerce as a groundbreaking technology that can set goals and make decisions. It can also act. In an ecommerce setting, that means performing tasks like proposing merchandising rule changes or generating reports on customer behavior patterns.
There’s huge potential for agentic commerce to dominate the way consumers research and buy products. As agentic systems become the expected digital entry point for shopping, hyperscalers like Google, OpenAI, and Anthropic are racing to own this space.
However, in the rush to create agentic commerce systems, these companies are missing an important ingredient that shapes the shopping experience — merchandising. While this technology will surely influence how people buy, it offers a massive opportunity to change how merchandising teams work. For agentic commerce to be part of the new AI-driven shopping process, it must augment and not replace human merchandisers.
Why Replacing Merchandisers is the Wrong Goal
Industry leaders in retail and tech tend to focus on a world where AI replaces human workers. The reality is that we still need humans involved in most aspects of business. Merchandising, like many things, is a process that requires human oversight. The narrative of “no-touch” AI breaks down fast in an enterprise setting for a few key reasons.
The Accountability Trap
Merchandising teams are accountable for tangible business outcomes, so if a leader delegates that work entirely to an AI, they are effectively taking that responsibility themselves. Most leaders are thoughtful about where they draw the line with AI — not because they’re unwilling to innovate, but because someone must ultimately be accountable for business outcomes.
As an old IBM training maxim puts it, “a computer can never be held accountable, therefore a computer must never make a management decision.” This highlights an enduring challenge in enterprise environments: tools should support decisions, but humans still own them.
Merchandising is Context, Not Just Execution
It takes human experience to translate ambiguous merchandising context into action. An AI doesn’t have the benefit of this experience. It won’t, for example, know that a specific supplier is facing delays or a competitor’s crop top just went viral on TikTok. Merchandisers provide the essential “why” behind the “what.” Without that human intuition to interpret the nuance of the business, AI won’t reach its full potential.
The Data Reality Check
Full AI autonomy is unrealistic given the chaotic state of enterprise data. Many companies struggle to implement clean product catalogs and event tracking. It is pollyannish to believe that we can suddenly feed an AI perfectly structured data on supply chains, business strategies, and external market trends. Until the data infrastructure exists to support it, the human merchandiser remains the critical bridge between data and decision.
The Real Opportunity Is in Augmenting Merchandisers
If we focus on using agentic software to augment merchandisers instead of trying to replace them, the problems around trust and accountability largely disappear. This approach gives leaders a way to elevate the expertise of their merchandisers.
It empowers workers, makes them more efficient, and reduces the amount of hiring required to support growth. An added bonus is that you don’t need to reinvent the wheel. Mapping software to existing organizational structures and workflows makes sense for enterprises that are already dealing with complex tech stacks.
Augmentation has the potential to reshape (in a good way) what a merchandiser’s week looks like. Merchandiser’s spend the bulk of their time focused on reactive and tactical work rather than strategy. Agentic systems can take on the tedious time-consuming tasks that eat up this valuable time. This includes things like reviewing reports or monitoring anomalies—important work that an agent can handle most of the time (no coffee break needed).
When AI can relieve people of these types of repetitive and mundane tasks, merchandisers are free to spend more time on intent and strategy. This is a tangible example of how technology supports the experts rather than removing them from the loop. Value comes from reclaimed focus and better decision-making.
A Practical Approach to Agentic Merchandising
Coveo’s approach to agentic merchandising centers on two types of agentic systems: agents and copilots. While these terms are often used loosely in the industry, we view them as specific product concepts that serve different needs for the modern merchandiser. Together, they act as a support system that allows teams to move away from thinking in terms of individual tools and start thinking in terms of intent.
Copilots: Turning Intent into Action
Copilots are essentially virtual assistants. They provide a natural language interface that fills the gaps where traditional user interfaces might feel slow or complex. A copilot helps a merchandiser move faster by turning a simple statement of intent into an exact change within the system. For example, a copilot can replicate a complex set of rules across dozens of international storefronts.
Agents: Taking on the Tactical Burden
Think of agents as specialists, virtual teammates that work in the background. They perform narrow, well-scoped tasks and handle the tedious work that can bog humans down. Your helpful virtual teammate is infinitely disciplined. It can check a report every hour or validate information across fifty catalogs to ensure every shoe, shirt, and belt buckle is categorized correctly.
An agent’s strength lies in investigating. It assembles evidence and proposes changes that human merchandisers can review and apply. They always require a final rubber stamp. This keeps your merchandisers involved and responsible for approval while the agent takes on the tactical burden of data analysis.
Beyond just performing tasks, copilots can help with decision support by “war-gaming” with different approaches to a problem, such as high return rates in a specific category. Copilots help teams avoid over-relying on familiar features and, over time, can orchestrate other agents, allowing a merchandiser to delegate an entire complex outcome to a specialized system.
Trust, Control, and Earned Autonomy
Coveo’s approach to agentic tools is very deliberately augmentation-first. We believe that everything the AI proposes must get final human oversight, at least at first. Agents and copilots can operate autonomously once they’ve earned the confidence to do so.
To make this work, we are applying enterprise-grade governance to these new agentic workflows. This includes investing in flows to review AI-proposed actions, reliable guardrails, and full auditability.
Merchandisers need to see the underlying data (e.g., underperforming SEO keywords, pre- and post-rule metrics, etc.) that an agent used to inform a decision or action before approving it.
Keeping human merchandisers in control, keeps systems in check. It ensures that no agent operates as a black box. This internal control is the best way to maintain a high standard of merchandising as we prepare for a reality where external agents begin to interact with the commerce ecosystem.
What Agentic Merchandising Really Means
For now, at least, AI lacks the essential business context and intuition that human merchandisers bring to the table. The shift to AI commerce — including AI merchandising — is more about using technology to help scale the type of rapid decision-making needed for enterprise commerce. These tools let merchandisers delegate tactical, repetitive, and tedious tasks to infinitely disciplined agents.
We think the future of commerce belongs to human-in-the-loop agentic systems and that the most successful teams will be those that use these high-powered tools to raise their floor of performance and elevate their strategic ceiling.
Human merchandisers provide value that AI agents can’t replicate. They understand the nuance involved with how consumers move through their individual buying journeys and the external factors at play. With the help of AI agents, merchandisers can now amplify this expertise. This combination of technology and human innovation is what will reinvent shopping experiences, ensuring they resonate with different types of shoppers who have many new ways to move through the buying journey.