In the evolving landscape of secure cloud automation, this paper introduces a novel AI-driven framework that integrates homomorphic encryption (HE), facial biometric authentication, and encrypted voice-command processing to achieve intelligent, privacy-preserving cloud resource management. Unlike prior models that depend on plaintext computation or modular scripts, our system performs encrypted biometric matching and command parsing, orchestrating cloud operations through secure tokens generated in the encrypted domain. This architecture, built atop AWS services such as EC2, Lambda, IAM, and Polly, ensures cryptographic isolation and integrity of sensitive workflows. Experimental results demonstrate authentication accuracy of 96.2%, command latency under 2.8 seconds, and significant performance gains over traditional interfaces, with zero leakage under chosen-ciphertext attack simulations. The proposed framework sets a new benchmark for secure, encrypted, and intelligent cloud management.