The increasing complexity of infrastructure design calls for innovative approaches that enhance efficiency and reduce human error. This study proposes an AI-driven methodology for Infrastructure Building Information Modeling (I-BIM), focused on the design of roadways and gravity-based utility networks. Leveraging Autodesk’s InfraWorks, Civil 3D, and Navisworks, a custom AI agent was developed to actively support users throughout the workflow. Unlike traditional chatbots, this agent is capable of autonomous reasoning and continuous self-improvement, offering real-time guidance and decision-making support. The research presents a comparative analysis between conventional manual workflows, basic conversational agents, and the proposed AI-enhanced solution. Results indicate that the integration of an intelligent agent can significantly streamline the design process, reduce repetitive tasks, and minimize the likelihood of errors. However, despite the promising outcomes, several methodological challenges were identified. These limitations highlight the need for further refinement of AI-agent frameworks to ensure their reliability, scalability, and effective integration into real-world infrastructure design environments.