This paper explores how agentic AI is reshaping network architecture—enabling both “AI for Networks” (automation, assurance, and self-healing) and “Networks for AI” (infrastructure optimized for distributed training, inference, and edge workloads). The paper outlines a high-level reference architecture for integrating AI agents with intent-based orchestration and network APIs, and proposes an Agentic AI Readiness Charter with practical principles for building deterministic, secure, and interoperable autonomy across open networking stacks.
