The collaboration frames agentic AI, OSS/BSS evolution and the knowledge plane as foundations for decision-safe autonomy
Communications service providers (CSPS) are gradually working to move from automation to autonomy, but this seachange is not simply a matter of adding more AI to existing operating models. It requires a deeper architectural and organizational shift from human-defined workflows to intent-based systems capable of planning, validating and executing network changes safely.
That is the focus of Ericsson and Telstra’s collaboration on autonomous networks. Announced in 2025, the partnership is designed to create, validate and industrialize business-driven use cases while addressing foundational challenges including fragmented data, the gap between business intent and network execution, multi-vendor complexity and AI trustworthiness.
Mats Karlsson, Ericsson’s Head of Solution Area Business & Operations Support Systems, said the collaboration is focused on both practical use cases and the foundation required for autonomy. “The central proposition is really that intent and the concept of knowledge [are] two fundamental aspects to going towards autonomous network,” he said.
That distinction is key. Automation, Karlsson said, still depends on people forming a step-by-step plan. “Automation is really still human-defined flows and human-defined configuration,” he said. “When you’re going towards autonomous network, you’re actually telling the network what do you want the network to achieve, what outcome, what is the intent.” In that model, agents create and execute the flows required to achieve the stated outcome.
For Telstra, the imperative is to capture the upside of autonomy without compromising network reliability. Mark Sanders, Telstra’s chief architect, framed the challenge in operational terms. “This is our network. This is our core business,” he said. “The opportunity is incredible…We’ve got to be able to do it with confidence.” The reason, he continued, is that AI agents are probabilistic, while network operations require deterministic confidence in the actions being taken.
This is where the knowledge plane becomes central. A co-authored Ericsson/Telstra industry perspective paper frames CSPs as caught in a “fragility trap”: the need to balance scale, reliability and efficiency can lead operators to minimize change, which then creates inertia. The proposed shift is toward intent-based autonomy, where CSPs define desired outcomes and the system autonomously plans, designs, validates and executes changes.
Sanders described the knowledge plane as the machine-readable context that makes that possible. Networks already produce large volumes of data, but autonomy depends on understanding the relationships between data, context, intent and actual network state. “Knowledge is the relationships,” he said. “But most importantly, we’ve coded those relationships…into the actual data…So now that the knowledge can be read by machines, they can reason over and they can act.”
Ericsson’s broader OSS/BSS strategy connects this autonomy vision to business execution. Its new OSS/BSS Business Value Pathways are designed to move CSPs beyond adding AI to legacy automation and toward specified transformation outcomes across monetization, experience and efficiency. The initial pathways include faster data transformation, zero-touch product launch, agentic AI service experience and agentic AI intelligent IT operations.
Karlsson emphasized that this cannot be delivered as a closed, one-off effort. “This will require a full industry collaboration,” he said. The path will not be a “big bang”; it will start with targeted use cases, expand the knowledge base over time and require vendors, operators and standards bodies to resolve conflicts between business intent, operational intent and network constraints.
The message for the industry is that autonomy is a new model for operational discipline. Engineers move from executing inside the loop to governing the loop. AI agents become useful only when grounded in intent, knowledge, policy and trust. And OSS/BSS becomes the control layer that helps CSPs sell, deliver and get paid in a more programmable, autonomous network future.
