5G Core Summit Blog

Summary

When the TM Forum first introduced its Autonomous Networks framework (with a scale from 0 to 5) in 2019, it was unclear how telecom operators could advance beyond Level 2 (partial autonomy) to Level 3 (conditional autonomy) or Level 4 (highly autonomous) in all but a handful of processes (e.g., fault management) and in a subset of their network estate (e.g., core network).

Fast forward to 2025, and many operators are now on the cusp of reaching Level 4, not for the entire network but for a growing proportion of it. Examples include telcos such as AIS Thailand, China Mobile, MTN South Africa, and STC in Saudi Arabia.

Agentic AI is a new technology that plays an increasingly important role in the journey to Level 4. Several of the aforementioned operators are applying AI agents to increase automation in areas such as complaint handling and alarm handling. However, agentic AI is not the answer to every automation challenge. It should be applied selectively and surgically, targeting dynamic domains where adaptability delivers real advantage. Furthermore, agentic AI must be built on a strong data foundation including domain-specific knowledge graphs.

One area to which agentic AI appears to be well suited is core networking. Although the core is not the biggest part of an operator’s CAPEX or OPEX, it plays an outsized role in the overall stability of the network. Core network reliability requires system resiliency that can cope with faults developing in individual network functions.

To enhance core network stability, operators should embrace a layered approach, using rules and scripts for routine tasks, and deploying agentic AI for more complex problems. Such problems can be solved using model self-reflection, evaluation, and chain-of-thought reasoning (empowered by the experience of domain experts). By doing so, operators can reshape the operations and maintenance (O&M) function of their core networks to achieve efficiency and high network stability, accelerating the evolution to Autonomous Networks Level 4.

Agentic AI will be key to achieving full network autonomy, but trust, explainability, and governance of AI is also critical. In addition, telcos must overcome the challenges of integrating AI with legacy systems and implementing it despite workforce skill shortages. By doing so, telecom operators can harness the real transformative potential of agentic AI without falling victim to marketing hype.

Autonomous Networks Level 4 is in sight

Autonomous telecom networks are key to operational efficiency and the pursuit of new revenue opportunities. Full autonomy (Level 5) remains a long-term aspiration, but Level 4—where human oversight is reduced to a minimum, and AI-driven automation governs network decisions—is the telecom industry’s target over the next five years.

Currently, fewer than half of telecom operators have moved beyond Level 2 automation, where AI is used primarily for monitoring and recommendations rather than autonomous decision-making. Even among leading operators, progress remains mixed. Some have successfully automated individual domains (e.g., core network), but true end-to-end automation remains rare.

According to the TM Forum’s definition, Level 4 networks self-adapt, self-optimize, and self-heal with minimal human intervention. To reach Level 4, the industry should take a phased approach:

• Between 2025 and 2027, telcos should expand AI-driven self-optimization in individual domains while improving their real-time intent-translation capabilities.
• Between 2028 and 2030, cross-domain closed-loop automation will start bridging radio access network (RAN), transport, and core networks, enabled by AI models that negotiate network resources dynamically.

A major characteristic of Level 4 network autonomy is the widespread application of intelligent agents. These agents can translate high-level business objectives into real-time network actions, underpinning the concept of intent-based automation. Moreover, as the name suggests, AI agents have agency: they can make their own decisions, thereby enabling closed-loop automation.

Agentic AI is a key enabler of Level 4

Agentic AI has sparked intense debate in the telecom industry and beyond. Some experts suggest that this is a new paradigm that will completely reshape system design; others say the probabilistic nature of large language models (LLMs) means this technology is inherently unreliable and hence unsuitable for mission-critical applications such as telecommunications.

In our view, agentic AI is not a panacea. For many service providers, the fundamental challenges of network operations—such as managing repetitive provisioning, monitoring faults, or orchestrating updates—are still best addressed through more traditional automation and deterministic rule-based systems. These approaches offer simplicity, reliability, and operational predictability for stable, clearly defined tasks where the processes rarely change and the risks of dynamic decision-making outweigh potential gains.

Agentic AI adds value in scenarios where adaptability and contextual reasoning are more important. In cases such as dynamic service assurance or automated root-cause analysis, agents can observe, plan, and act semi-autonomously in response to both customer intentions and emergent network behavior.

No agentic architecture can succeed without a foundation of high-quality, trusted, and normalized data. Data readiness is an absolute prerequisite, since AI agents need unified, real-time visibility spanning OSS, BSS, and IT ecosystems. This must be supported by robust data observability, governance, and security frameworks.

The potential of agentic AI lies in combining reasoning, memory, perception, and goal-directed planning to create intelligent systems that exhibit cognitive behavior that aligns with business objectives. For this vision to work, agents must be based on well-defined ontologies and share declarative interfaces, cross-agent knowledge, and memory pools.

Applying agentic AI to improve core network reliability

With continuous network evolution, the core network becomes more complex, and the number of managed objects increases. Signaling errors in the packet core are common and can have a severe impact on the end service. Therefore, automating the core network is key to ensuring overall resilience and reliability.

With today’s cloud-based core networks, the automation solutions are often fragmented. Operators choose from a range of tools for each level of their network stack (infrastructure, container as a service, network functions, services, etc.). As a result, they often end up with many siloed automation tools.

As cloud-native technologies advance and new services emerge, core networks have become more complex. Increasingly, the O&M teams of the operators are unable to keep up with their growing workloads with their existing, siloed tools. To address these challenges, operators should address the automation of the core network holistically. Additionally, they should consider the use of intelligent agents to make core network O&M simpler and more reliable.

For example, a monitoring and troubleshooting process typically involves three personas:

• Monitoring engineer
• Core network engineer
• Decision maker (manager)

With O&M agents, activities traditionally carried out by monitoring and O&M engineers can now be automated. This doesn’t obviate the need for engineers; instead, it frees up their time from mundane tasks so they can focus on higher-value activities such as design and planning. Additionally, the decision maker benefits by receiving inputs more quickly and, potentially, more reliably.

For example, the production processes of China Mobile across several provinces use AI agents in its core network intelligent O&M controller. These agents have enhanced the efficiency of handling complaint and alarm tickets by over 60% and improved analysis accuracy to over 90%.

Appendix

Roberto Kompany, Principal Analyst, Mobile Infrastructureaskananalyst@omdia.com

Citation Policy

Request external citation and usage of Omdia research and data via citations@omdia.com.

Omdia Consulting

Omdia is a market-leading data, research, and consulting business focused on helping digital service providers, technology companies, and enterprise decision makers thrive in the connected digital economy. Through our global base of analysts, we offer expert analysis and strategic insight across the IT, telecoms, and media industries.

We create business advantage for our customers by providing actionable insight to support business planning, product development, and go-to-market initiatives.

Our unique combination of authoritative data, market analysis, and vertical industry expertise is designed to empower decision-making, helping our clients profit from new technologies and capitalize on evolving business models.

Omdia is part of Informa TechTarget, a B2B information services business serving the technology, media, and telecoms sector. The Informa group is listed on the London Stock Exchange.

We hope that this analysis will help you make informed and imaginative business decisions. If you have further requirements, Omdia’s consulting team may be able to help your company identify future trends and opportunities.

Get in touch

www.omdia.com

askananalyst@omdia.com

Copyright notice and disclaimer

The Omdia research, data, and information referenced herein (the “Omdia Materials”) are the copyrighted property of TechTarget, Inc. and its subsidiaries or affiliates (together “Informa TechTarget”) or its third-party data providers and represent data, research, opinions, or viewpoints published by Informa TechTarget and are not representations of fact.

The Omdia Materials reflect information and opinions from the original publication date and not from the date of this document. The information and opinions expressed in the Omdia Materials are subject to change without notice, and Informa TechTarget does not have any duty or responsibility to update the Omdia Materials or this publication as a result.

Omdia Materials are delivered on an “as-is” and “as-available” basis. No representation or warranty, express or implied, is made as to the fairness, accuracy, completeness, or correctness of the information, opinions, and conclusions contained in Omdia Materials.

To the maximum extent permitted by law, Informa TechTarget and its affiliates, officers, directors, employees, agents, and third-party data providers disclaim any liability (including, without limitation, any liability arising from fault or negligence) as to the accuracy or completeness or use of the Omdia Materials. Informa TechTarget will not, under any circumstance whatsoever, be liable for any trading, investment, commercial, or other decisions based on or made in reliance of the Omdia Materials.