This chapter explores the shift from passive adoption of generative artificial intelligence (GenAI) tools toward educator-led design of AI-powered solutions. While recent advances have made it easier than ever to create custom instructional tools using natural language interfaces, many technologies remain disconnected from classroom realities. Past experiences with educational technology have shown that tools developed externally, without educator input, often fail to meet the diverse and evolving needs of students and instructors. With GenAI, the opportunity now exists for educators to create personalized materials, assessments, scaffolds, and tutors aligned with their learning objectives. However, risks related to bias, oversimplification, misinformation, and data privacy remain significant—particularly when relying on commercial platforms governed by proprietary models and changing terms of service. To address these risks, Arizona State University has developed the CreateAI platform: a secure, model-agnostic environment built within the university’s infrastructure. CreateAI supports over 40 large language models and allows users to compare, build or customize, and deploy tools without sharing data externally. Educators can co-design AI companions and learning experiences tailored to their students, with embedded ethical safeguards and institutional alignment. This approach presents a sustainable, flexible model for institutions seeking to ensure that AI integration enhances—rather than replaces—the role of the educator.