InfraGenie - Automating the process of generating a deploy ready Infrastructure as Code (IaC) using AI

Abstract

Infrastructure as Code (IaC) has become integral to modern cloud deployment, yet automating its generation remains a complex and manual task. This paper presents InfraGenie, a novel multi-agentic system designed to automate the generation, validation, and deployment of Terraform-based IaC configurations for Azure, specifically tailored for e-commerce applications. The system orchestrates specialized agents Requirement Extraction, Terraform Generation, Validation, Review, and Pricing to iteratively refine infrastructure configurations based on user inputs, security policies, and scalability constraints. Extensive experimentation was conducted to evaluate various training approaches, revealing the limitations of single-model strategies and guiding the adoption of the proposed agent-based architecture. Functional and non-functional testing, along with expert evaluations, confirm the system’s accuracy, usability, and real-world applicability. This work contributes to the body of knowledge by demonstrating that a coordinated agent-based pipeline can address the reliability, compliance, and semantic alignment challenges often found in automated IaC generation. InfraGenie offers a foundation for scalable, secure, and intelligent infrastructure provisioning in cloud-native development.