Photo by Kindel Media on Pexels

Ericsson Redraws the Telecom Operations Map with Agentic AI at Its Center

In what could prove to be one of the most consequential architectural shifts in telecom operations in years, Ericsson has formalized a new blueprint for its Operations Support Systems and Business Support Systems (OSS/BSS) stack — one that treats AI agents not as peripheral tools but as first-class citizens of the entire software framework. The Swedish networking giant is calling it an “agentic service experience layer,” and industry observers are already taking note of its potential to redefine how carriers manage increasingly complex 5G and multi-access edge computing environments.

The announcement represents a significant evolution from earlier generations of AI-assisted network management, where machine learning models were largely siloed, reactive, and required heavy human supervision. What Ericsson is proposing is fundamentally different: autonomous, goal-oriented AI agents that can reason, plan, and execute across multiple domains of a carrier’s operational and commercial infrastructure simultaneously.

What Is an “Agentic Service Experience Layer” — and Why Does It Matter?

At its core, the concept of agentic AI moves beyond traditional automation scripts or even conventional machine learning pipelines. An AI agent, in this context, is a software entity capable of perceiving its environment, setting objectives, breaking down complex tasks, and taking sequential actions to achieve outcomes — all with minimal human intervention.

Ericsson’s architectural blueprint embeds these agents directly into the OSS/BSS stack, creating a dedicated layer that sits between raw network data and the business logic that operators rely on for service assurance, customer experience management, billing, and network slicing orchestration. Rather than waiting for human operators to trigger workflows, these agents can proactively identify degraded service conditions, reallocate network resources, update charging rules, and even flag potential revenue leakage — all in near-real time.

For telecom operators running large-scale 5G Standalone (SA) networks with dynamic slicing requirements, this kind of autonomous intelligence is increasingly not a luxury but a necessity. The sheer volume of telemetry data generated by a modern RAN, combined with the complexity of service-level agreements across enterprise and consumer segments, has long since exceeded what human-driven operations can efficiently handle.

Technical Architecture: Agents Across the Full OSS/BSS Spectrum

Ericsson’s model appears to distribute AI agents across several functional domains within the OSS/BSS stack. On the operations side, agents are designed to handle fault management, performance optimization, and configuration management tasks that traditionally required dedicated network operations center (NOC) personnel. On the business support side, agents integrate with charging, policy, and customer management functions to enable more dynamic, context-aware service delivery.

Critically, the architecture supports multi-agent orchestration, meaning individual specialized agents can collaborate — passing context, negotiating priorities, and escalating to human operators only when genuinely ambiguous decisions arise. This is a meaningful technical distinction from simpler automation platforms and aligns with broader industry momentum around the concept of intent-based networking, where operators define desired outcomes and the network figures out how to achieve them.

The blueprint also incorporates large language model (LLM) capabilities as a reasoning engine for certain agent types, enabling natural-language interfaces for network engineers who may need to query complex operational states or instruct agents to execute non-standard remediation strategies without writing custom code.

Industry Context: The Race to Autonomous Networks

Ericsson’s move does not occur in a vacuum. The broader telecom industry has been inching toward what the TM Forum and major vendors define as Autonomous Network levels 4 and 5 — representing conditional and full self-management, respectively. Nokia, Huawei, and a growing cohort of cloud-native BSS vendors have similarly been investing heavily in AI-driven automation. But formalizing an agentic layer within the OSS/BSS architecture as a defined, productized component marks a degree of structural commitment that goes beyond roadmap promises.

For operators, the business case is increasingly clear. Labor costs associated with manual network operations remain stubbornly high, particularly as networks grow more heterogeneous with the proliferation of Open RAN, private 5G, and fixed-wireless access deployments. McKinsey and other analysts have estimated that AI-driven automation could reduce network operating expenses by 20 to 40 percent over the next five years for carriers that successfully deploy it at scale.

Operator Readiness: The Human Factor

Despite the technological promise, the transition to agentic OSS/BSS will not be without friction. Operators must contend with legacy system integration challenges, data quality issues that can undermine agent decision-making, and significant organizational change management. Ericsson and its peers will need to provide robust explainability tools so that network engineers can understand and trust the actions taken by autonomous agents — especially in high-stakes scenarios involving service outages or regulatory compliance.

Outlook: A New Operational Paradigm Is Taking Shape

Ericsson’s architectural bet on agentic AI signals that the era of purely reactive, human-supervised telecom operations is drawing to a close. As 5G Advanced matures and the industry begins early planning for 6G, the ability to manage hyper-dense, software-defined networks at machine speed will increasingly separate competitive operators from those struggling to maintain margins.

The formalization of an agentic service experience layer may well become the template that defines OSS/BSS design for the next decade — and with Ericsson putting its full architectural weight behind it, the rest of the industry will be watching closely to see whether the promise translates into measurable operational outcomes for the carriers bold enough to deploy it first.