Foundational Design Principles and Patterns for Building Robust and Adaptive GenAI-Native Systems

Explore foundational design principles and architectural patterns for developing reliable GenAI-native systems in this 36-minute conference presentation from Onward! 2025. Learn how to address the major challenges of unpredictability and inefficiency in generative AI by integrating cognitive capabilities with traditional software engineering principles. Discover five key pillars for GenAI-native design—reliability, excellence, evolvability, self-reliance, and assurance—and examine architectural patterns including GenAI-native cells, organic substrates, and programmable routers that guide the creation of resilient and self-evolving systems. Understand the components of a GenAI-native software stack and analyze the technical, user adoption, economic, and legal implications of these emerging systems. Gain insights into a paradigm shift that advocates for robust, adaptive, and efficient GenAI-empowered systems through this research presentation by Frederik Vandeputte from Nokia Bell Labs, which aims to inspire future research and provide a conceptual framework for the development of next-generation AI systems.