• USD/RUB USD/RUB 84.90
  • EUR/RUB EUR/RUB 94.97
  • CNY/RUB CNY/RUB 12.67
  • Bitcoin Bitcoin BTC $83034
  • Ethereum Ethereum ETH $2501
  • Ripple Ripple XRP $1.39
  • Solana Solana SOL $109.40
  • Cardano Cardano ADA $0.25
  • USD/RUB USD/RUB 84.90
  • EUR/RUB EUR/RUB 94.97
  • CNY/RUB CNY/RUB 12.67
  • Bitcoin Bitcoin BTC $83034
  • Ethereum Ethereum ETH $2501
  • Ripple Ripple XRP $1.39
  • Solana Solana SOL $109.40
  • Cardano Cardano ADA $0.25

AI instead of agencies: How an FMCG company generates 1,000 cards per hour

AI instead of agencies: How an FMCG company generates 1,000 cards per hour
Most Popular
07.10
Marketplaces will be allowed to public procurement of schools and kindergartens from October 2026
07.10
Technological fee for electronics: rates from 25 to 500 rubles from December 2026
07.10
Poizon enters the Russian market with a commission of up to 29% for sellers
07.10
Ozon increased the FBO insurance rate to 0.0838% per day
07.10
Kadalinsky Industrial Park on the border with China is put into operation
07.10
Kazakhstan deepens Aktau port: cargo traffic along the Middle Corridor will grow by 30%
The AI system from FOKINA.AI allowed the FMCG company to abandon weeks of manual work and create up to 1,000 product cards per hour. This made it possible to reach marketplaces faster, reduce costs and increase conversions.

With the rapid growth of e-commerce and fierce competition between sellers on marketplaces, the speed of new product withdrawal is becoming a crucial success factor. One of the largest players in the FMCG sector faced the need to multiply the pace of product card preparation and solved the problem using the AI Product Visuals AI platform from FOKINA.AI .

The task was ambitious — to generate more than 1,000 cards per hour without the involvement of contractors, while maintaining control over the quality of visual content. In the traditional scheme, creating cards takes a long time: several rounds of edits, photographing, adapting to each site, and coordinating between departments. All this stretched the time-to-market for weeks, reducing revenue and slowing product turnover.

The solution included a generator that creates a ready-made card based on a basic image and a text query: with photos, short video previews, texts and automatic adaptation to the standards of the largest marketplaces (Ozon, Wildberries, Yandex Market, etc.). The architecture is based on custom versions of Stable Diffusion and JuggernautXL models, as well as its own pipelines with a module for editing visual, infographics and resizing.

The technical requirements included working within the limits of the security and confidentiality policy, with rights to all created materials, as well as the possibility of full management without external contractors.

The development took 5 months. The first stage involves analyzing more than 10,000 old flashcards to highlight the characteristics of the ideal content. Next— parallel AI setup and infrastructure implementation, followed by trial operation. The result was a scalable solution that eliminated the company's dependence on agencies, reduced content production costs by 73%, and accelerated the launch of new products by 44%.

In addition to reducing costs and time, the company has achieved growth rates: the conversion rate on the card has increased by almost 50%. This has become a critical advantage in an environment where 71% of sellers are confident that visual content directly affects sales. Unlike template solutions, AI Product Visuals ensures the individuality of the cards, their adaptation to the sites and compliance with brand standards.

This case clearly shows: AI in e-commerce is no longer an experiment, but a tool that gives a competitive advantage.