Heima Claims Breakthrough in AI Agent Security

Heima's approach to securing AI agents leans on two core technologies: non-custodial agent smart accounts and TEE-Secured Execution. The non-custodial accounts allow AI agents to execute transactions independently without requiring third-party custody of user funds, which mitigates the risk of hacks and asset mismanagement. For protecting sensitive data and the agent's operational logic, Heima employs Trusted Execution Environments (TEEs), which are secure and isolated hardware areas within a processor. This move into AI agent security is a strategic evolution for the project, which was formerly known as Litentry. Founded by ex-Parity software engineer Hanwen Cheng, Litentry's initial focus was on decentralized identity (DID) and cross-chain identity aggregation. The rebranding to Heima signals a broader ambition to provide a foundational infrastructure layer for what it terms a "verifiable agentic economy." The project's foundation was supported by a seed round in October 2020, which included investors like FBG Capital, Candaq, Hypersphere Ventures, and Signum Capital. This early backing was for its initial vision as Litentry, which aimed to build a DID protocol on the Polkadot network. Heima's focus on TEEs places it in a growing field of projects aiming to provide secure and private computation for AI on the blockchain. Competitors in this space include Phala Network and Oasis Network, which also leverage TEEs to enable confidential smart contracts and AI model execution. The overarching goal for these platforms is to solve the trust issue with AI agents, ensuring they operate as intended without tampering and that sensitive data remains private. The use of TEEs allows for cryptographic proof of an agent's correct execution, which is a critical component for their adoption in high-stakes financial applications.