In fact, Wildberries (as part of the merged RVB company) is doing what has long been the "bottleneck" of large marketplaces: reviews help to sell, but their volume turns into information noise. The new feature takes fresh comments and collects a summary of frequently repeated characteristics, with the text automatically updated as new reviews become available. If the product has different variations — size, color, and package — the retelling is formed separately for each, which reduces the risk of "mixing" impressions from different versions.
The key point for the trust of buyers and sellers is that retelling does not replace the original source. RVB emphasizes that the "Important of the reviews" does not affect the rating and does not hide the reviews themselves — it is the navigation of meaning, not a new rating system. Moreover, users have a feedback tool.: they can check whether the final text is correct and report any inaccuracies. This is an important safety net for any AI reports where generalizations, loss of context, or controversial formulations are possible.
It is worth evaluating the business effect separately. For a marketplace, such a "quick read" layer usually increases the conversion rate to an order (a person quickly understands what is most often praised and scolded for) and at the same time reduces returns: the buyer relies less on the product card and more on the experience of others. For sellers, this means a new front of reputational work: it is important not just to collect reviews, but to achieve repeatable formulations about real advantages (seam quality, fabric density, packaging, size matching). It is the repeatability and frequency that become the "fuel" for the summary.
At the same time, there are risks typical for large platforms. If unscrupulous participants start massively "hammering in" the same theses, they may try to influence the content of the retelling. Therefore, anti-fraud filters, anomaly detection, and the weight of "trusted" reviews are especially important. Outwardly, the feature looks like a convenience for the buyer, but in fact it pushes the market towards more mature rules: less spam — more verifiable specifics.
RVB directly explains why the understanding of reviews needs to be accelerated, and makes this the central idea of the launch.
"Reviews are an important factor in making a purchase decision, most buyers pay attention to them when choosing a product. Neural network retellings are a new format for interacting with reviews, which allows you to quickly highlight the main thing without having to read dozens or hundreds of opinions. This is another step towards making online shopping convenient and intuitive," said Polina Ovchinnikova, Head of Product Reviews, Questions and Rating at RVB.
Testing, according to the company, started in the spring of 2025 on a small group of users and received positive feedback, and now the feature is available to most users of the mobile application and should appear on the website.
