Constructor has launched a new AI agent for eCommerce merchandising teams, as retailers look for faster ways to diagnose and adjust online product discovery.
The product, called Merchant Intelligence Agent (MIA), sits inside Constructor’s search and product discovery platform. It uses a conversational interface and in-product prompts to explain why items appear where they do in search results, category pages, and campaigns.
Retailers already use Constructor for onsite search and discovery across online storefronts. The supplier lists customers including The Very Group, Sephora, White Stuff and Under Armour.
Merchandising workflow
Merchandising teams manage the rules, campaigns, and visibility decisions that shape what shoppers see across a retailer’s digital properties. When something changes unexpectedly, teams often have to use multiple tools and dashboards to work out what happened.
Common scenarios include products dropping out of a key search query, promoted items losing prominence, or a campaign failing to deliver expected visibility. In these cases, teams may need to check inventory data from upstream systems, review configuration rules, and examine signals from ranking models and shopper behaviour.
MIA is designed to support that investigative work. It provides AI-generated explanations inside the Constructor platform, including prompts that surface without a user query. It also offers a chat-based interface that accepts natural-language questions about product visibility and performance.
Constructor positions MIA as a way for merchandisers to ask direct questions about search and discovery outcomes-for example, why a product is not appearing for a query, what factors influence its ranking, and what changes might improve visibility for a category or campaign.
Explanations and actions
The agent has three functions: explaining results, suggesting next steps, and automating selected actions. It generates explanations based on a retailer’s store data, ranking logic, and merchandising rules.
MIA produces summaries of search results and product rankings and links them to merchandising performance. It also offers recommendations tied to team goals, such as improving ranking for a key query or increasing visibility for a campaign.
The release also adds AI-assisted campaign creation, with MIA able to pre-fill campaign set-up fields. Merchandisers can review and edit inputs before deployment.
For internal reporting, MIA can generate executive-ready summaries of trends and performance metrics for sharing across an organisation.
Reasoning engine
Constructor links MIA to its Commerce Reasoning Engine, described as the logic layer that determines what the platform shows a shopper and when. The engine processes real-time shopper behaviour and context and blends that with historical interactions, survey responses, offline signals, and merchandising configurations.
MIA acts as an explainability layer for those decisions, with the aim of giving teams responsible for search and discovery outcomes greater transparency.
MIA joins Constructor’s existing AI agent products, including AI Shopping Agent and AI Product Insights Agent, which are positioned as shopper-facing tools for conversational questions, recommendations, and product information.
Retail AI push
The launch comes as eCommerce teams test generative AI beyond customer service. Product discovery is a key commercial lever, and small changes in ranking and availability can affect conversion and revenue. Merchandising teams often run frequent experiments with boosts, rules, and promotions.
At the same time, eCommerce stacks have grown more complex. Retailers often run separate systems for inventory, pricing, promotions, analytics, and personalisation. That fragmentation can slow diagnosis when results diverge from expectations.
Constructor’s approach focuses on packaging diagnostics into a tool that fits day-to-day workflows, alongside automation features that turn recommendations into campaign set-up steps.
In a statement, Eli Finkelshteyn, CEO and Co-founder of Constructor, said:
“Merchandisers are experts at retail strategy. They shouldn’t have to waste their time doing menial investigations that AI can do for them. If a merchandiser has a great idea, we want to let them have AI do the execution. If a merchandiser gets a question about product visibility or campaign performance from their boss, MIA can give them the answer. If a launch happens and something doesn’t look right, a merchandiser often doesn’t have the time to investigate – they need answers now. MIA is here to help. And as Constructor’s dashboard expands with more and more functionality, merchandisers shouldn’t have to memorise where everything is. They should just be able to ask MIA.”
Constructor is demonstrating MIA at Shoptalk Spring through booth activities and a product showcase session led by Finkelshteyn and Andrew Carcova, VP of Global Solution Engineering at Constructor.
