NGMN Urges Industry Alignment to Harness Agentic AI for Autonomous Mobile Networks

The Next Generation Mobile Networks (NGMN) Alliance has released a significant publication, 'Network Automation and Autonomy Phase III: Agentic AI for Autonomous Mobile Networks,' which outlines the motivation, opportunities, and requirements for integrating Agentic AI into autonomous mobile networks. This report provides a pragmatic reference framework for Agentic AI-driven network automation, assesses ecosystem maturity and gaps, and offers considerations for accelerating safe, interoperable, and scalable deployments. It emphasizes that Agentic AI is poised to enable trusted Level 4 autonomous mobile networks, characterized by automated decision-making, self-optimization, self-healing, and self-management. This development is critical for network practitioners, particularly those in mobile network operations, because it moves the conversation beyond basic automation to a vision of truly autonomous systems. The shift from isolated proof-of-concepts to operational deployments of Agentic AI, prioritizing high-value use cases, means that network engineers and architects will soon be dealing with agents managing complex workflows across multiple network domains. The report highlights that the success of commercial-scale adoption hinges not just on AI model capabilities, but on comprehensive support systems, ecosystem alignment, and fragmentation mitigation. This directly impacts how MNOs plan their architectural transformations, operational models, and governance frameworks. The broader trend in cloud, DevOps, and AI has consistently moved towards greater automation and intelligence. From Infrastructure as Code (IaC) and GitOps in cloud environments to AIOps platforms for IT operations, the goal has always been to reduce manual intervention and increase system resilience and efficiency. Agentic AI represents an evolution of this trend, building on the foundations laid by earlier forms of network automation and AIOps. While previous iterations focused on data analysis, prediction, and closed-loop optimization, Agentic AI introduces systems capable of reasoning, planning, collaborating, and executing actions autonomously. This aligns with the long-term vision of Intent-Based Networking (IBN), where networks understand and adapt to high-level business intents rather than requiring granular, manual configurations. The NGMN's call for ecosystem alignment echoes similar efforts in Kubernetes and other open-source projects, where community collaboration is vital for standardizing interfaces and ensuring interoperability. In practice, this means network professionals should begin to familiarize themselves with the principles of Agentic AI and its implications for network architecture. They should advocate for open standards and interoperable solutions within their organizations and across the industry to prevent vendor lock-in and ensure future flexibility. Practitioners will need to develop new skill sets related to AI governance, policy adherence, observability, explainability, and operational safety, as these are identified as crucial support systems for Agentic AI deployment. Furthermore, the report suggests a focus on well-controlled, high-value use cases initially, implying that a phased, strategic approach to Agentic AI adoption will be key. This is not just about implementing new tools, but about a fundamental transformation of network operations, requiring a shift in mindset towards managing intelligent, autonomous systems rather than just configuring devices.