The multinational company, spanning personal health management, medical imaging, and patient monitoring systems, is implementing a top-down AI capability restructuring plan. Departing from traditional AI deployment models limited to technical teams, Philips has for the first time incorporated large model application skills into its all-employee professional competency framework. This decision directly mirrors the global AI industry’s shift from “expert-driven” to “universal empowerment.”
“In healthcare, AI tools must break out of the algorithm lab,” revealed Patrick Mans, head of data science at Philips. The company has established a tiered empowerment system: management undergoes initial training on GPT-like large models, while an enterprise-level AI sandbox environment is opened for employee experimentation. Notably, the training curriculum places special emphasis on chip-level computing power allocation principles within medical scenarios, laying a technical foundation for subsequent workflow automation.
In its implementation strategy, the company demonstrates the cautious approach characteristic of healthcare AI. By launching an internal innovation challenge, Philips prioritizes collecting AI application proposals for low-risk scenarios. This gradual path not only builds user confidence but also systematically verifies performance under different computing configurations. Currently, 37% of clinical support processes have introduced AI agents, primarily handling non-diagnostic clerical tasks.
“The true value of medical AI lies in freeing up professional manpower,” Mans pointed out a key industry pain point: physicians spend an average of 30% of their working hours on administrative duties. Philips is testing an intelligent document system that combines NLP large models with the heterogeneous computing power of medical-specific chips, with the potential to reduce that time to under 10%. This efficiency gain, rooted in infrastructure reconstruction, could redefine the healthcare service value chain.