Perceptron's CEO says its 800,000-node AI data network has been "a decentralized brain connected to a centralized memory." Today it's moving the memory.
The network's nodes are people sharing idle bandwidth to gather AI training data. On Thursday it announced Walrus, the Sui-based storage network, as its primary storage layer, replacing centralized storage. Every dataset gets a verifiable ID, so buyers like hedge funds and AI labs can check the chain of custody themselves.
It's also the backend for Perceptron's upcoming Data Questing Platform, where contributors annotate and label data for $PERC and reputation points.
The CEO calls it a shift "from trust to verify yourself." Which is a polite way of saying the old setup was trust us.
Source: https://walrus.xyz/blog/perceptron-walrus-verifiable-data-provenance/