NEAR Protocol Launches Infrastructure for AI Agents
The NEAR Foundation has launched what it calls "confidential cross-chain infrastructure" designed for the agentic economy. The platform aims to provide a secure runtime environment and economic layer for autonomous AI agents. This marks NEAR's evolution from a blockchain to an integrated execution layer for decentralized AI.
- NEAR co-founder Illia Polosukhin was a co-creator of the Transformer architecture at Google, the foundational technology behind modern LLMs like ChatGPT, before starting NEAR. - The new infrastructure includes "IronClaw," an open-source AI agent runtime that uses Trusted Execution Environments (TEEs) to run agents in encrypted hardware enclaves, protecting their data and credentials. - A core component is "NEAR Intents," a system that allows a user or an agent to specify a desired outcome (e.g., "swap 100k USDC to BTC"), letting the protocol handle the complex cross-chain routing and execution across more than 35 blockchains. - To support the ecosystem, the NEAR Foundation has launched a $20 million AI Agent Fund to invest in agent tokens and decentralized AI applications being built on the protocol. - The platform introduces "Confidential Intents," which enables private transactions where balances and activity are not publicly visible on-chain, a feature designed for AI agents handling sensitive financial strategies. - To address the computational needs of AI, NEAR launched a Confidential GPU Marketplace, a network where AI workloads involving sensitive data can be processed without exposing the data to the GPU operators. - The initiative aims to create an "agentic economy" where autonomous AI agents can act as economic participants, earning and spending tokens by completing tasks on a decentralized marketplace. - This technology is bundled for consumers into the Near.com "super app," which is designed to abstract away blockchain complexities like gas fees and private keys for the end-user.
