Oro is a Bittensor subnet (sn15) focused on evaluating AI-powered shopping agents in real-world commerce scenarios. Developers (miners) submit Python-based agents that search, compare, and recommend products from a dataset of 2.5 million real items. Validators run these agents in isolated Docker environments, scoring their performance on accuracy and compliance using the ShoppingBench benchmark. The top-performing agents are rewarded with TAO emissions, incentivizing the creation of effective AI shopping assistants. Oro aims to set a gold standard for AI agent evaluation in online commerce, providing open participation and transparent incentives.