It’s Monday morning. Before the first meeting, a campaign manager needs a boost rule applied across twelve locales. A buyer has flagged that a top category is underperforming and wants answers by noon. There’s a weekly performance report that takes three hours to pull together manually. And somewhere in the middle of all that, there’s a seasonal strategy that needs attention before the window closes.
By Friday afternoon, the tactical work got done. The strategic work did not.
This is the reality for most merchandising teams. Not because they lack expertise or ambition but because their tools are built for execution, not intent.
Coveo’s Merchandising Agent is built to change that.
Meet the Coveo Merchandising Agent
The Merchandising Agent is a natural language AI companion that lives inside the Coveo Merchandising Hub, accessible as a side panel or full screen. It knows where you are in the product and adjusts accordingly. A merchandiser working on a product listing page gets a different context than one reviewing search analytics. The conversation starts where the work is.
The way it works is straightforward. A merchandiser describes what they need in plain language; “show me the queries with the most traffic and the lowest conversion rate” or “boost the shoes category across all my European locales” and the agent gets to work. It surfaces insights, prepares actions, and presents everything for review before anything goes live.
It is worth being clear about what the Merchandising Agent is not. It is not a replacement for the Coveo Merchandising Hub or the expertise of the people using it. It does not make changes autonomously. Every action it recommends requires explicit approval from a merchandiser before it is applied. The intent is to move faster and think bigger, not to hand over control.
What Merchandisers are Doing With the Merchandising Agent Right Now
Analytics and Reporting
Every reporting interface is built around the questions its designers anticipated. The obvious ones get made easy. The less obvious ones, the cross-cuts, the edge cases, the questions that only come up when something feels off, often require a workaround, or just don’t get asked at all.
The merchandising agent changes that. Because it works in natural language against the full data set, it handles the long tail of analytical questions that standard reporting interfaces were never built to answer. A merchandiser who wants to understand which queries are generating high traffic but low add-to-cart rates across a specific product category, filtered by a recent date range, can just ask. The answer comes back in minutes, structured and ready to act on.
Teams are using the agent to run performance reviews, delegate weekly reporting tasks, and surface trends in customer search behavior that would have taken hours to piece together manually. It became the most common starting point for teams using the Merchandising Agent and the reason is simple. The questions merchandisers actually have do not always map neatly to the reports already built for them. The Merchandising Agent fills that gap.
Rule Creation
If reporting and analytics is where teams start, rule creation is where they feel the biggest impact.
A retailer with a catalog spanning thirty locales needs to run a promotional boost across every market. The products differ by region. The category names differ too. Mapping the right products to the right rules, validating the data, and applying the configurations manually is a project that takes most of a day.
With the merchandising agent, a merchandiser describes the goal: find the equivalent of the shoes category across every locale in my catalog, validate it against real performance data, and create the boost rules. The Merchandising Agent works through the catalog, surfaces the plan for review, and applies it once approved. A day-long task becomes a five-minute one.
This is where the agent earns its place most clearly. Bulk actions, cross-catalog changes, and any task that involves doing the same thing across a large number of configurations are exactly the kind of work it handles well. For experienced merchandisers who know precisely what they want to accomplish, the merchandising agent removes the friction of getting there. The judgment stays with the merchandiser. The execution gets handled.
Decision Support
A category manager notices that returns are running higher than usual in a specific product line. She knows something is wrong but is not sure whether the issue is in the product listings, the search rankings, the facet configuration, or something else entirely.
Working through those possibilities used to mean pulling data from multiple places, forming a hypothesis, and testing it manually. The manager describes the problem and asks for help thinking it through. The Merchandising Agent surfaces relevant data, suggests possible interventions, and helps structure a plan. The decision still belongs to the merchandiser. The agent makes it faster and better informed.
Teams using the agent for decision support report that it surfaces options they would not have considered first. Acceptance rates on the agent suggestions are high, which reflects the quality of what it produces when given a real problem to work through.
A Rising Tide for the Whole Team
One of the less obvious benefits of the merchandising agent is what it does for team consistency.
In any merchandising organization, not every team member operates at the same level. A senior merchandiser with years of category experience thinks and works differently from someone who joined last month. The gap between them is not just skill but accumulated knowledge about how the business works, what the data means, and how to approach a problem.
The Merchandising Agent helps close that gap. When best practices get encoded into how the team uses the agent, how to analyze a category, how to contextualize a performance drop, how to structure a promotional rule, that knowledge becomes accessible to everyone. A newer merchandiser working through an unfamiliar problem can follow the same analytical approach that a senior colleague would use. The floor of the whole team rises.
This is not about replacing expertise. It is about making sure that expertise reaches further.
Built With Safety and Control in Mind
The merchandising agent can handle the vast majority of actions available inside the Merchandising Hub. Two categories are intentionally excluded: deletions and filter rules.
Deletions are excluded because they are difficult to reverse and easy to regret. Filter rules are excluded for a more specific reason. If a filter rule references a product field with inconsistent data coverage, it can break search results in ways that are hard to trace. A rule that looks correct in the merchandising agent could silently cause problems in production. Keeping filter rule creation outside the agent’s scope eliminates that risk while teams get comfortable with the tool.
Beyond these boundaries, every session the Merchandising Agent runs is logged for audit and review. Merchandisers can see what it recommended, what was approved, and what was applied. Nothing happens without a paper trail, and nothing goes live without a human sign-off.
Control and accountability stay exactly where they belong, with your team.
Where This Is Going
The immediate value of the merchandising agent is in time savings and task delegation. The longer-term direction is more significant.
The next phase of development focuses on expanding the depth and sophistication of the questions the agent can answer, alongside delivering richer analytical context and next-best actions. Instead of only responding to the questions a merchandiser thinks to ask, the agent will proactively surface what deserves attention and recommend what to do about it. A merchandiser reviewing a category page will not just get data. They will get a recommendation, grounded in that data, with a clear path to acting on it.
After that, the roadmap moves toward scheduled and trigger-based flows. A performance report that currently depends on someone remembering to run it every morning becomes something the merchandising agent handles automatically, delivering the summary before the workday starts. A content gap analysis that happens quarterly because it takes a week to prepare could run on a continuous basis instead.
Further out, the vision is semi-autonomous and fully autonomous agents that manage entire workflows, with human oversight applied where it matters most and removed where it adds friction without adding value.
The goal through all of this is for the merchandising agent to become the place where merchandising work begins. Not a feature inside the Merchandising Hub, but the starting point that connects strategy to execution across the entire commerce stack. Merchandisers are experts and the tools they use should reflect that, the agent is how we are getting there.
Ready to See It in Action?
The Merchandising Agent is live now inside the Coveo Merchandising Hub, included as part of your existing subscription. Reach out to your customer success manager or request a demo to see what it can do for your team.