Ozon has launched pilot testing of its own AI assistant designed to guide users from "just browsing" to placing an order. The tool, which operates on anonymized platform data, is currently available only to a limited group of customers, but the very fact of its launch marks a new stage in the competitive race among Russian marketplaces.
This is a logical continuation of the trend: the largest market players are actively integrating conversational services directly into the sales funnel, aiming to reduce selection time and, consequently, boost conversion. We are witnessing a fundamental shift from passive search to active consultation.
How the Ozon assistant works
At the current stage, the assistant targets users who have not yet decided on a purchase. After switching to a special mode, the user enters a chat where the AI asks clarifying questions: about budget, key product characteristics, and the specific task for which the purchase is being made. Based on these answers, the system generates a personalized selection, taking into account a range of critically important parameters:
- Product price.
- Delivery time and cost.
- Seller rating.
- Reviews from other customers.
The assistant's operation is based on the analysis of Ozon's anonymized data: search queries, purchase history, a vast array of reviews, and current prices. This approach allows the model not just to output random items but to simulate the logic of an experienced consultant.
What this means for the market
The conversational format fundamentally changes the search paradigm itself. Instead of tedious scrolling through product cards and adjusting filters, the user simply describes their need in words. The system translates this request into a specific selection, bringing the interaction closer to consulting with a live salesperson rather than a classic search results page. For the platform, this is a powerful retention tool: it allows them to "not let go" of an indecisive shopper beyond the service, gently guiding them toward placing an order.
Limited access at the pilot stage is standard practice. This allows the company to test the accuracy and relevance of recommendations, the system's resilience under conversational load, and the usability of the interface itself before scaling the solution to the entire audience.
Notably, at the end of June this year, the main competitor—the merged company Wildberries & Russ—announced the launch of a similar service. Their assistant, integrated into the mobile app, also helps select products, including gifts, and compare similar items. Wildberries plans to teach the AI to suggest products based on photos and account for purchase history, integrating the development into the overall strategy of embedding AI across all ecosystem projects.
My take on the situation: The synchronized move by the two giants is not a coincidence but a marker of market maturity. Competition is shifting from "who has more products" to "who better understands the customer." Success here will be determined not so much by the model's technological sophistication as by the quality of the data it is trained on. And here, the advantage will go to those who can build the most seamless and trustworthy dialogue with the user. Pilot launches are just the first step in this long game, where the stakes are consumer loyalty and wallet share.