As artificial intelligence (AI) advances across domains, its adoption in high-stakes environments such as industrial predictive maintenance requires trust, transparency, and usability. Ensuring AI-driven decisions are explainable and user-centered is essential for real-world deployment. While existing AI frameworks offer valuable principles and tools, integrating them effectively into development practices needs further development. This study evaluates two established AI frameworks, Schneiderman’s Human-Centered AI (HCAI) and DARPA’s Explainable AI (XAI), assessing their suitability for predictive maintenance from a user-centered perspective. HCAI provides valuable principles for human-AI collaboration, trust, and ethical oversight but requires more structured implementation guidance. Conversely, XAI offers robust explainability techniques but does not fully address iterative refinements, participatory design, or structured governance. To address these limitations, this paper presents the Methodological Human-Centered Agile Framework for AI-XAI Development, integrating HCAI and XAI principles into a structured methodology. Rather than treating these frameworks as separate, the proposed approach enhances their integration by combining structured governance, business understanding, data-driven insights, and Agile’s iterative refinements. This ensures AI development aligns with industrial needs, regulatory compliance, and evolving operational demands. By embedding continuous user feedback, participatory design elements, and fairness evaluations, this framework provides a systematic process for integrating explainability and human-centered principles into AI systems. Beyond predictive maintenance, it offers a scalable, governance-ready approach for embedding explainable and ethical AI into industrial systems. Future research should explore how this framework can be refined for broader applications, including its adaptability across various industries and sustained mechanisms for explainability and user trust.