High-tech systems are growing more complex due to mass customization, integration of diverse technologies, and long lifecycle demands. Customers increasingly expect service contracts based on performance and availability, yet diagnostics remain largely reactive and reliant on human expertise. This position paper proposes a Pervasive Intelligent Diagnostics (PID) framework that integrates pervasive sensing, model-based digital twins, and hybrid AI for predictive diagnostics and sustainable lifecycle management. PID embeds collaborative sensing and reasoning within operational environments. We outline a research agenda for leveraging pervasive sensing and digital twins to advance intelligent diagnostics in high-tech systems. Key directions include: integrating heterogeneous sensor data with system models, automatically generating diagnostic models, and evaluating them in high-tech case studies. Expected benefits include reduced downtime, improved resource use, and stronger retention of expert knowledge. These outcomes align with industry roadmaps for sustainable, dependable systems.