Covalent: The structure, scalability, and AI applicability of a decentralized data pipeline @Covalent_HQ is a decentralized data infrastructure with a modular pipeline structure for on-chain data indexing, designed to ensure data integrity and operator honesty through cryptographic verification at each stage. This structure is divided into Block Data Ingestion, Indexing/Processing, Storage, Query Serving, and Auditing, with proofs generated at every stage recorded on the ProofChain. In the first stage, Block Data Ingestion, Block Specimen Producers extract raw data from supported blockchains and convert it into 'Block Specimen' form. This data is stored on IPFS, and the corresponding hash is posted to the ProofChain to preserve it in an immutable state. Subsequently, Block Result Producers process the Block Specimen into structured data to generate 'Block Result', which is also stored on IPFS, and the proof value is recorded on the ProofChain. Finally, Query Node Operators retrieve this Block Result to handle API queries, with this process also aiming for complete decentralization in the future. All operations are verified by Auditors, which serves as the basis for reward distribution. In terms of latency, there is an average delay of two blocks. This design choice prioritizes the completeness of cryptographic verification over real-time performance, which limits ultra-low latency applications while maintaining data reliability. The modularization of each stage allows for parallel processing, contributing to scalability, but the structural characteristics of using ProofChain and IPFS inevitably introduce temporal overhead due to verification and network synchronization. Regarding scalability, Covalent supports over 100 blockchains, with the number of BSPs and BRPs continuously increasing. However, if the number of BSPs outpaces the growth rate of BRPs, processing bottlenecks may occur, and the rising costs associated with large data storage and synchronization delays of new nodes can act as potential constraints. Currently, the query nodes are centralized, necessitating a complete transition to decentralization, and the complexity of verification coordination among auditors may increase with network expansion. In terms of AI and agent applications, Covalent's structured multi-chain data is advantageous for reliable learning and analytical data. Notably, the AI Agent SDK, released at the end of 2024, supports agents in reading and writing on-chain data and performing verification automatically. Additionally, the EWM (Ethereum Wayback Machine) can serve as a permanent and verifiable data repository for AI model training and regulatory compliance. However, the two-block delay and response speed limitations due to the verification process may pose constraints for real-time inference AI applications, and there is a risk of increased response times as query complexity and data volume grow. The operator incentive structure must remain continuously aligned to maintain network trust and data freshness. Key observation metrics include the number of block specimens and results generated, the active counts of BSPs, BRPs, and QNOs, the staking ratio of CXT, the incidence rate of data discrepancies, query processing latency, and the history of reward distribution. Alongside this, off-chain metrics such as the number of paid API customers, expansion of partnerships, increase in supported chains, and growth rate of the operator community are also considered key indicators of the project's actual growth and adoption. As of Q4 2024, there are 24 active BSPs and 15 BRPs, with Block Specimen generation increasing by 14.5% compared to the previous quarter. The staking ratio of CXT relative to circulating supply has risen to 32%, and the number of EWM light client operators is around 458 based on the testnet. There are over 70 paid API customers, many of whom are linked to major DeFi and AI projects. In summary, Covalent is building a decentralized data trust infrastructure through a multi-layered structure that separates data collection, verification, storage, querying, and auditing, while gradually strengthening the foundation for AI and agent application expansion. The future transition to EWM-ProofChain based on Cosmos SDK is expected to be a key turning point for throughput expansion and bottleneck resolution.