Sahara Ai has been very popular recently and has already been arranged on Binance. Many friends are already aware of it, but there is also an AI project called @ codata_io quietly rising in the same track recently. The same thing is that both are working on Web3 data infrastructure, but their directions are completely different. Sahara is more like a decentralized data marketplace, solving the problem of "where to find data and how to use it", mainly serving data access and permission control for AI projects, while Codatta is more focused on "what does this data really mean". It does tag annotation of on chain behavior to build structured metadata, such as whether this address is a robot, what type of transaction it belongs to, and whether this person has been listed. Speaking in human language means that the former makes data available, while the latter makes data understandable, with completely different positioning, and can even be used together in the future. The core mechanism of Codatta is staking as confidence, which simply means that whoever thinks the label is reliable can pledge and endorse it, and the community reaches consensus through a voting mechanism. Tags can be generated by AI or manually supplemented by humans, such as labeling addresses, transaction behaviors, and NFT projects, ultimately forming a semantic layer of data network. These data can not only serve on chain analysis and anti witch mechanisms, but also feed AI models as training sets or be directly used in DeFi project recommendations and risk management. Overall, Codatta is doing the thing of "data meaning", making pure data on chain behavior contextual, semantic, and structured. A relatively complete data contribution and application loop has been constructed using decentralized annotation+staking mechanism+AI verification+reward distribution.