AI Agent Finds Hidden Marketplace Seller Losses in 15 Minutes
Data Power has developed an AI agent that automatically identifies hard-to-spot profit losses among marketplace sellers.
According to the developer’s estimate, a large seller can lose as much as 1.5 million rubles per week due to calculation errors and blind spots in analytics.
The system connects to the seller’s account via API and analyzes sales, advertising and inventory levels. It identifies campaigns where promotional spending has exceeded the margin, detects declining conversion rates for product listings and flags products on which advertising budgets are being spent without a noticeable return.
The calculations and anomaly detection are performed by software code based on predefined rules, while the neural network only turns the results into conclusions and recommendations for the manager. The developers say this approach is intended to prevent situations in which the model independently “fills in” financial figures.
A manual search for such deviations takes an analyst about eight hours, while the agent takes roughly 15 minutes. The final decision remains with a human. The service is designed for sellers with monthly revenue of at least 20 million rubles; the data is anonymized and stored on servers in Russia.
