Static data contracts in IoT systems are inadequate for dynamic, behavior-driven environments where real-time adaptation is required to maintain data quality, responsiveness, and resource efficiency. This paper is presented as a visionary architecture paper to outline foundational directions for future AI-integrated data governance in adaptive IoT systems. It introduces two core components: (1) an AI Engine responsible for analyzing behavioral data and predicting system conditions such as user presence, preferences, or load patterns, and (2) a Generative AI Engine based on Large Language Models (LLMs), which automatically produces adaptive data contracts based on insights derived from the AI Engine. These contracts define expected data schema, validation constraints, and quality expectations. The architecture is motivated by smart environment use cases, such as museums, where adaptation to user behavior can improve experience, efficiency, and personalization. We present five integrated layers: IoT Data Collection, Data Processing, AI Prediction, Generative AI Contract Generation, and Contract Enforcement with Feedback. Each layer contributes to enabling runtime adaptation in data-intensive IoT systems. This early-stage architecture sets the foundation for future implementations that aim to automate data governance using AI-driven decision logic and generation mechanisms.

Conference Title

European Conference on Software Architecture (ECSA) 2025

Conference Country

Greece

Conference Date

Sept. 15, 2025 - Sept. 19, 2025

Conference Sponsor

Springer