I saw an emerging AI data company these days, two post-00s @ pencermatega and @ Carlos Georgescu, which valued the company at 21.5 billion RMB in less than a year. The company is called AfterQuery @ AfterQuery, founded by a high school acquaintance who interned at Meta together. The company was founded in February 2025 and was selected for the YC accelerator without even having a product. The founder initially wanted to become a financial AI agent, but found that the model performed poorly on real professional workflows, so he turned to "encoding how experts think and work into training data". YC Winter 2025 batch, established around the beginning of 2025. The core products include SFT (with thought chain), expert rubric's RL, API/MCP agent environment, computer usage trajectory, etc., covering fields such as finance, law, healthcare, engineering, etc. Claiming to serve almost all cutting-edge laboratories in the United States, and also collaborating with some Chinese laboratories; Customers include Nvidia (Nemotron), Thinking Machines Lab, Legora, and others. The expert network claims to have nearly 100000 verified in-service professionals. In this field, there were startups such as Scale AI and Surge AI with valuations of billions of dollars, followed by AI companies like AfterQuery with valuations of billions of dollars. The traditional model of crowdsourcing and expert networks combines a lot of related hardware and systems for data collection and evaluation, as well as the construction of various AI training environments. There are also companies like @ axisrobotics that specialize in simulation and first person perspective mobile app task data in the field of robotics, and they can continue to participate in web3 crowdsourcing models https://s.kaito.ai/dywfbXk ）Sahara AI (SAHARA) is a decentralized data annotation/data service platform (DSP) that allows users to earn tokens or stablecoins by performing tasks such as image annotation, audio transcription, and text evaluation. The token will be listed on TGE on June 26, 2025. @BuildOnSapien (SAPIEN) is positioned more purely as a "decentralized AI data factory", with global contributors annotating and verifying data quality using Proof of Quality. The token will be TGE listed on Base on August 20, 2025, and subsequently listed on Binance and other platforms. At present, the market value in web3 is not very large. One of the core reasons is that there is no real data collection and sales model formed after the coin is issued, only relying on token speculation. This model is just a small pool of mutual swaps, and it is impossible to grow into a project with a valuation of billions of dollars. Can the latest @ axisrobotics robot data protocol break free from this pure speculation mode without a closed loop of revenue. Data is still the most cash flow project in the AI field, and there are also many projects. I am optimistic about the development of this field.