Tinkoff has developed the algorithmic cashback technology with recommendation models - Tinkoff RECO, and launched its pilot. The technology is created for selecting an individual cashback for customers for goods, brands, purchases in stores and in different product categories.

The representatives of the company notes that AI algorithms have been trained using the purchases of 8 million customers of the company for 2 years in various categories (FMCG, restaurants, appliances and electronics, clothing, etc.).

The technology may also be useful for business partners. It will help a store or manufacturing brands determine which products to sell, whether to launch a product or brand offline, which customer needs to be offered a new category of products, and what cashback should be offered to clients.

Tinkoff RECO technology is a new solution developed as a part of ReceiptNLP service for decrypting text information from receipts using neural networks.

Contacts: 

Vera Leychenko, Head of Clearing and Settlement Products Department, Tinkoff

Original (in Russian)