Bitrecs
Bitrecs(subnet 122) is a protocol built on the Bittensor network, specializing in product recommendations for e-commerce websites. Its core function is to use simple rules such as {1，2，3} >= {1，2} to recommend available product collections (stock-keeping units) to online shoppers; these recommendations appear in common sections on product pages such as “Similar to This” or “You May Also Like.” The network leverages an alliance of large language model (LLM) calls initiated by miners to make the “best guess” about what customers may be interested in. Miners receive queries containing shopper context, such as products viewed, cart contents, or browsing history, and use prompting techniques, such as “recommend products complementary to a given product,” to create personalized suggestions. Validators then evaluate these responses based on criteria including relevance, diversity, latency, and potential to increase conversion rates, selecting the best recommendations. Feedback loops from real user interactions refine the system over time, improving accuracy and increasing average order value (AOV）。 Bitrecs operates entirely on an opt-in basis, allowing merchants to easily adopt a simple plugin to enable subnet inference directly on their product pages for predictions. This decentralized approach ensures scalable,AI-driven recommendations without relying on centralized data silos.
Overview
Bitrecs(subnet 122) is a protocol built on the Bittensor network, specializing in product recommendations for e-commerce websites. Its core function is to use simple rules such as {1，2，3} >= {1，2} to recommend available product collections (stock-keeping units) to online shoppers; these recommendations appear in common sections on product pages such as “Similar to This” or “You May Also Like.” The network leverages an alliance of large language model (LLM) calls initiated by miners to make the “best guess” about what customers may be interested in. Miners receive queries containing shopper context, such as products viewed, cart contents, or browsing history, and use prompting techniques, such as “recommend products complementary to a given product,” to create personalized suggestions. Validators then evaluate these responses based on criteria including relevance, diversity, latency, and potential to increase conversion rates, selecting the best recommendations. Feedback loops from real user interactions refine the system over time, improving accuracy and increasing average order value (AOV）。 Bitrecs operates entirely on an opt-in basis, allowing merchants to easily adopt a simple plugin to enable subnet inference directly on their product pages for predictions. This decentralized approach ensures scalable,AI-driven recommendations without relying on centralized data silos.