Pulumi AI leverages large language models to simplify the discovery, learning, and utilization of cloud infrastructure APIs, acting as a guide through the dynamic cloud landscape by providing tailored solutions. Recent enhancements include integrating Pulumi Package schema data, which improves the accuracy and relevance of generated Pulumi programs by overcoming limitations such as outdated provider knowledge and the risk of generating non-existent APIs. Despite the challenge of schema data size exceeding the capacity of current generative AI context windows, a new approach was devised to select and inject relevant schema information, enhancing Pulumi AI's accuracy and responsiveness without increasing latency. These improvements allow Pulumi AI to generate more precise infrastructure as code solutions, such as creating EKS clusters or deploying static websites on Azure, and reference up-to-date resources like Google AlloyDB. As Pulumi AI continues to evolve, it aims to empower users through features like interactive documentation and shareable conversation links, contributing to the broader field of AI research and cloud engineering.