The AI-Native Manifesto for Large-Scale Agile Software Development

Summary

The paper introduces the AI-Native Large-Scale Agile Software Development Manifesto, a proposal for reorganizing large software-development work when AI, particularly large language models, participates as a first-class actor. It argues that existing large-scale agile frameworks remain heavily human-centric and manual, depending on coordination meetings, synchronized artifacts, and role-based handoffs that can limit real-time adaptation. The manifesto is organized around six principles: parallel processes, intent-driven teams, living knowledge, verification-first assurance, orchestrated agent workforces, and reusable blueprints. Together, these principles are intended to replace meeting-driven, document-heavy, sequential workflows with a more intelligent, adaptive, and continuously learning system. The paper positions the manifesto as an organizational response to advances in AI-assisted software engineering, while its stated contribution is a set of values and principles rather than a new model, implementation, or empirical benchmark. It is classified under software engineering and artificial intelligence on arXiv.