Wireless capsule endoscopy (WCE) enables screening of the gastrointestinal tract by a swallowable imaging system. However, contemporary WCE systems have several limitations, which often result in low diagnostic yield. This paper introduces the concept of a next generation WCE system with embedded intelligence aiming to effectively minimize diagnostic errors. The proposed system is based on a novel wirelessly-powered hardware-software architecture integrating reconfigurable components that are optimized in terms of area-time complexity and power consumption. It integrates multispectral and 3D vision modules, and embedded intelligence for video quality control, accurate localization of the capsule and automatic detection of a broad spectrum of abnormalities. The feasibility of the proposed WCE system is qualitatively assessed with respect to the results obtained, and novel research directions are drawn.