Welcome to Sapiens

This litepaper presents the architecture and technical implementation of a decentralized Fully Homomorphic Encryption (FHE) AI protocol and vector database designed for Large Language Models (LLMs). Our approach leverages blockchain technology to ensure data integrity and immutability, adapts current blockchains to operate under FHE for secure and efficient consensus, and utilizes emerging decentralized GPU compute solutions to enhance computational scalability and performance. We detail the challenges faced, our innovative solutions, and the technical methodologies employed to create a secure, scalable, and privacy-preserving data processing infrastructure.