Ericsson says operators exploring autonomous networks need accurate, continuously updated inventory before AI agents can make operational decisions. The company also says fragmented or outdated service and resource inventories can limit closed-loop automation.
Why inventory is the input layer
Ericsson says AI agents rely on inventory data to understand which network resources exist, how they are configured, and how services depend on them. It says incorrect inventory can drive service provisioning on unavailable resources, wrong fault diagnosis, or capacity planning based on outdated information.
Ericsson also says stale topology can lead to incorrect service-level calculations or a digital twin that represents a network state that no longer exists. In its framing, inventory has to stay synchronized with changes across the network if it is going to support autonomous operations.
What the inventory has to cover
Network inventory records physical, logical, and virtual resources, including sites, nodes, links, network functions, spectrum, network slices, configurations, service states, and the relationships among them. Ericsson says inventory data can be collected from network management systems, controllers, orchestration platforms, and assurance systems before being reconciled into a common operational model.
Radio access, transport, IP, optical, cloud, and core systems can expose operational information using different models, so inventory platforms have to normalize that data before it is used across network domains. Nokia has identified multi-vendor support, network discovery, reconciliation, and topology management as requirements for inventory systems used in autonomous networks.
Inventory is not the whole decision stack
Ericsson says a unified inventory does not contain everything needed for autonomous decision-making. Its model adds intent, policies, constraints, service and customer context, historical evidence, and domain rules.
Ericsson describes a knowledge plane as the layer that combines those sources using shared semantics and ontologies. TM Forum describes the same kind of layer as one that turns fragmented network data into information supporting intent-driven operations and cross-domain autonomy.
Digital twins and conflict control
Digital twins provide virtual representations that can model network behavior and assess changes before those changes reach production. ETSI is working on network digital-twin specifications covering standardized architectures, data models, interfaces, testing, and closed-loop verification across multi-vendor environments.
Ericsson says a digital twin built on outdated inventory can represent a network state that no longer exists and can affect planning and automated network operations. TM Forum’s July Catalyst looked at agents, applications, orchestrators, and service functions operating over shared infrastructure, including cases where an energy-saving action shifted traffic onto a neighboring cell and where a new business slice competed with an existing priority commitment.
Read for autonomous-network toolchains
The practical read for operators and suppliers is that inventory quality is not a housekeeping issue once AI agents are expected to act on the network. Ericsson’s model links inventory with intent management, closed-loop automation, network data, and knowledge models, so the data layer becomes part of the control stack.
For vendors building autonomous-network platforms, the benchmark is not just a database that holds resource records. It is whether the system can keep those records current, reconcile multiple domains, and feed decision logic that can detect conflicts before actions hit production.