Sixth generation networks will integrate native AI and multi-agent architectures to achieve autonomous management that interprets intentions, coordinates resources, and adapts operations in real-time through a semantic control plane that orchestrates distributed agents between edge and core. Recent proposals outline a framework in which agents based on large language models (LLM) operate as bounded reasoning entities governed by policies, combining deterministic infrastructure with semantic abstractions of intention and context, to execute network decisions in a distributed and secure manner. This vision is contextualised within governance and network orchestration frameworks that seek to continuously close feedback loops in response to variations in demand, topology, and operational policies.
Traffic and complexity projections support the transition to 6G: estimates of AI impact on the network management layer indicate increases in demand for AI capabilities for operations and optimisation, and visions of business scenarios highlight operational benefits today in RAN management and slice orchestration, with the aim of advancing towards semi-autonomous services and real-time resource negotiations. In this regard, industry projections indicate that global WAN traffic could multiply by between 3 and 7 times by 2034 compared to 2023, driving autonomous management solutions that reduce costs and response times in the face of demand spikes and service failures. This momentum is aligned with the transition to AI-driven network environments, where slice orchestration, monitoring, and resource trading are executed by agents acting collaboratively.
In terms of architecture, the proposed solutions for 6G contemplate an agent skirt that operates across multiple domains: device, edge, and core, with a hierarchical reasoning chain that breaks down intentions into feasible actions and integrates them with domain specialists to generate optimal network configurations, demonstrating viability in IBN (Intent-Based Networking) implementation plans supported by agents and multi-agent reasoning. These approaches emphasize the need for an autogenic control plane, capable of translating business intentions into operational policies, SLA verification, and executing changes in network slices and in the management of RAN and CN resources.
The standardization frameworks and regulatory perspectives highlight the importance of an autonomous approach governed by AI, with approaches already being discussed in forums such as TM Forum, 3GPP, and ETSI to consolidate autogenic management and the use of AGI in networks. In view, these works underscore the need for cognitive reasoning, security verification, and compatibility across domains to achieve a safe and scalable migration to efficient and resilient 6G networks, with capabilities for self-configuration, self-optimization, and self-recovery in complex scenarios.
1. Agentic AI for 6G networks: autonomous control plane for slice orchestration and monitoring - arXiv - 07/2026
2. From Agentic to Autogenic Network Management for AI-Native 6G and Beyond: A Standards Perspective - arXiv - 07/2026
3. The Dawn of 6G: Empowering a User-Centric Ecosystem with Agentic AI - ZTE Magazine - 02/2026