Grass Network: Building a Decentralized AI Data Layer

Grass is a decentralized network scraping platform deployed on the Solana chain, combining AI, Depin, and Solana technology. Its positioning is as the data layer for AI, aiming to assist companies and organizations in training AI models by utilizing unused internet bandwidth. Grass achieves web scraping through a browser extension application, allowing users to share idle bandwidth resources and be rewarded with Grass Points. Currently, the network has over 2 million users running nodes, scraping a large amount of data for AI models.

Technical Architecture

Grass has built a sovereign data Rollup network on Solana, capable of handling the entire process from data sources to processing, verification, and building datasets. The network mainly consists of the following components:

1. Validator (: Receives, verifies, and batches Web transactions from routers, generating ZK proofs to validate on-chain session data.

2. Router ): Connects Grass nodes and validators, relays bandwidth, and reports network metrics.

3. Grass Node ( Node ): Use the unused bandwidth of users to capture public Web data, and node operators are compensated based on the amount of relayed data.

4. ZK Processor: Validity proof of batch processing Web request session data, submitted to the L1 blockchain.

5. Grass Data Ledger: Store the fetched data and link the data to the chain for proof.

6. Edge embedding model: Transform unstructured Web data into structured data that meets AI model requirements.

Technical Features

The Grass network is located between the client and the Web server, handling the routing of Web requests and data scraping. Its main technical features include:

1. Grass Data Ledger serves as permanent storage, recording the metadata and lineage of data sets.

2. The ZK processor records the data set source, generates proofs, and submits them in batches to L1, solving the high throughput problem.

3. Metadata records the source website and routing nodes of the dataset, implementing a mechanism for rewarding nodes based on their contributions.

4. The incentive mechanism encourages users from popular regions to join, expanding the network scale and data storage capacity.

Node Operation and Security Mechanism

Grass nodes run for free and are primarily used to relay Web requests and return encrypted responses. Node operators are rewarded based on relay traffic, which is allocated according to reputation scores and geographic demand. Node reputation scores include aspects such as integrity, consistency, timeliness, and availability.

In terms of security mechanisms, Grass only routes internet traffic and does not access users' personal data. The network uses bandwidth encryption to protect users and collaborates with AppEsteem for 24-hour security monitoring.

Grass Token Features

Grass token holders can participate in the network in the following ways:

1. Transactions and Repurchases: Support for network scraping transactions, dataset purchases, etc.

2. Staking and Rewards: Stake to the router to promote network traffic and earn rewards.

3. Network Governance: Participate in proposals and voting to promote network development.

Currently, the annual yield of Grass staking is about 45%, with approximately 33% of the tokens participating in staking, and the total amount exceeds 26 million.

Router Staking

Users can stake Grass to earn rewards through the Router, with different commissions for different Routers. Currently, the staking amount for DBunker is about 1.43 million, with a minimum staking period of 7 days and a 10% commission. Users can connect their wallets to stake Grass and earn rewards.

Summary

Grass is committed to building a fair and open Decentralization Data Layer to address the ethical and quality issues of internet data extraction. Its unique technical architecture and incentive mechanism provide AI companies and protocols with a raw source of decentralized data, showing promising development prospects in the intersection of web3 and AI.