At the Agentic Core Summit held on March 3 during MWC Barcelona 2026, Shen Chengying, President of MAE Domain, Huawei Cloud Core Network Product Line, delivered a keynote entitled "From Scenario-Specific Autonomy to Single-Domain Autonomy: Advancing Core Networks to AN L4", announcing the official release of the core network AN L4 target solution CORESpirit. He emphasized that core network AN L4 must prioritize two strategic goals: high stability and high efficiency. The implementation will follow a two-phase roadmap: the first phase focuses on achieving autonomy in high-value scenarios, specifically fault management and change management, while the second phase aims for comprehensive single-domain autonomy across all scenarios, including network planning, construction, maintenance, optimization, and operations.
Shen Chengying, President of MAE Domain, Huawei Cloud Core Network Product Line, delivering a keynote
According to Informa's latest industry survey, operators are increasingly turning to AI to overcome core network O&M challenges, with AN L4 emerging as the definitive path forward. By leveraging cutting-edge AI technologies, including large language models (LLMs) and agents, the industry aims to achieve end-to-end (E2E) automated closed-loop management within specific domains and scenarios.
Huawei's ICNMaster prioritizes high stability and high efficiency by spearheading the industry's first AN L4 target solution—CORESpirit(Network agent of the core network). Accompanied by a two-phase evolution roadmap, the plan begins in Phase 1 (2025–2027) with the launch of a 'one-stop' intelligent interactive portal and the mastery of autonomy in high-value fault and change management. Moving into Phase 2 (2027–2029), CORESpirit will expand to deliver all-scenario collaborative autonomy within a single domain of the core network.
Phase 1 of AN L4: Scenario-Specific Autonomy
One-Stop Intelligent Interaction Portal: Powered by guided multi-turn dialogues and long-term memory, NOEMate accurately interprets user intent with a remarkable 90% accuracy rate. Utilizing role-based and preference-aware algorithms, it delivers tailored responses for diverse user profiles. Furthermore, the portal streamlines operations by intelligently aggregating data, alarms, and resources—automatically invoking tools and flexibly generating reports. This integration boosts O&M efficiency by over 30%.
Fault Scenario Autonomy: CORESpirit-Fault Management automatically responds to and handles faults based on their severity. For minor faults, it delivers root cause analysis along with recovery recommendations. For major faults, it intelligently selects available paths to preferentially restore services. This autonomy significantly reduces average service recovery time—from 90 minutes down to just 15 minutes.
Change Scenario Autonomy: CORESpirit-Change Management auto-generates configuration commands by leveraging LLMs and verifies the commands within the digital twin network. This streamlined process cuts service rollout time from hours to mere minutes.
The fault scenario autonomy has already been commercially deployed across 14 operators' networks worldwide, boosting troubleshooting efficiency by more than 60% and enabling fast service recovery within 15 minutes.
Phase 2 of AN L4: Core-Network Single-Domain Autonomy
CORESpirit has evolved from providing scenario-specific autonomy—whether fault or change management—to delivering layered autonomy across network planning, construction, maintenance, optimization, and operations. Guided by user service intent, it coordinates multiple scenarios and automatically sustains service objectives. By adopting the digital twin technology built on the interaction-mechanism model, CORESpirit enables interactive self-learning with physical networks. Leveraging the large agent model (LAM), it gathers environment feedback, continuously aligns with the target objectives during the feedback process, and independently completes E2E tasks. In this way, a fully autonomous core network is realized—capable of self-generation, self-repair, and self-optimization.
Huawei aims to collaborate with global operators and industry organizations to advance the development of AN L4 standards and drive commercial deployment, while continuing to explore AI applications in core network O&M to enhance reliability and efficiency—ultimately helping operators accelerate digital and intelligent O&M transformation.