Pond is a decentralized AI model layer that focuses on modeling and predicting on-chain data using graph neural networks (GNNs). By analyzing interactions between accounts and contracts, it provides crypto-native AI models for scenarios such as DeFi risk management, MEV strategies, security detection, and asset recommendations. Developers can access prediction results through its open model interface, or participate in collaborative model creation and ownership sharing, lowering the barrier for Web3 applications to integrate AI.