EigenCloud has launched EigenAI and EigenCompute on mainnet alpha, marking a significant step in bringing verifiable computing to the artificial intelligence sector. These services are designed to address the trust deficit in AI by ensuring that AI operations are transparent and secure, akin to the guarantees provided by smart contracts, according to EigenCloud.
Artificial Intelligence (AI) has been transformative across various sectors, yet it faces a persistent trust issue. The opaque nature of AI models, which often operate as ‘black boxes’, leaves users uncertain about the integrity of AI inputs and outputs. This is particularly concerning in high-stakes scenarios, such as financial technology (DeFi), where verifiable integrity is crucial.
EigenCloud aims to resolve these concerns with EigenAI and EigenCompute, which provide verifiable results from AI agents. These agents are subject to cryptoeconomic incentives, including penalties for failing to deliver verifiable outcomes. This framework allows for confident decision-making using AI, as emphasized by Sreeram Kannan, CEO of EigenCloud.
EigenAI offers a verifiable LLM inference API to eliminate the risks associated with unverifiable AI services. By ensuring that AI prompts and responses are unaltered, EigenAI makes AI reliable for critical applications. It supports a range of open-source LLMs and provides an OpenAI-compatible API, facilitating easy integration for developers.
EigenCompute allows developers to execute complex AI agent logic securely offchain while maintaining onchain security standards. It uses Trusted Execution Environments (TEEs) initially, with plans to expand to other verifiability methods like cryptoeconomic security and ZK proofs. This ensures that offchain computations are secure and verifiable.
EigenCloud’s initiatives have attracted collaborations with major industry players like Google and Coinbase. These partnerships aim to standardize agent-to-agent interactions and create a robust framework for verifiable AI applications. The integration with Google, for instance, focuses on extending blockchain-powered verifiability to the Agents Payment Protocol (AP2).
Developers are invited to build applications using EigenCompute, which supports immediate integration with EigenAI for LLM inference. The alpha phase will focus on user feedback to refine the services, with plans to secure additional resources to backstop applications during the transition to general availability.
As EigenCloud continues to develop its verifiable AI infrastructure, the company positions itself as a pivotal player in the evolution of secure and transparent AI applications, addressing both current limitations and future demands in AI technology.
