Ericsson is moving AI agents into the heart of telecom operations. Its new blueprint treats them as core parts of OSS/BSS. The shift marks a new phase for AI-native telecom systems.
Rather than adding generative AI tools to older workflows, Ericsson wants agents guiding decisions. These agents can plan, reason, test actions, and execute tasks. The model connects network operations with customer, billing, and service teams. It also fits AI-native 5G and future 6G planning.
At the center sits an agentic service experience layer. It aims to coordinate customer journeys, service lifecycles, and assurance. That could help operators launch offers faster. It may also reduce manual work across separate systems.
For example, a team could state one business goal in plain language. Ericsson uses the example “reduce churn among high-value customers”. The system would then create catalog items, charging rules, and provisioning steps. That reduces handoffs between product, billing, and network teams.
The architecture relies on several specialist agents. Experience agents study customer journeys and service quality. Revenue agents work on pricing, upsell, and margin targets. Network agents connect with automation tools for topology and resource changes.
This approach may appeal to operators under pressure to move faster. Many still run older support systems and complex workflows. Replacing everything at once rarely works. Ericsson instead pitches agents as a layer above existing domains.
However, the model needs trusted data to work well. Bad data could drive bad automated decisions at scale. Ericsson links the framework with its Telco DataOps Platform. This streams customer, service, and network data into one pipeline.
The company also extends closed-loop management beyond network alarms. Agents can watch experience indicators, such as NPS and service quality. They can adjust offers or settings, then measure business impact. That moves customer experience from reporting toward active management.
This move also sharpens vendor rivalry. Nokia recently advanced its own AI agent strategy for network operations. Operators like Verizon have also built agentic AI systems internally. Ericsson offers a packaged path for carriers lacking large AI engineering teams.
Still, operators will examine cloud choices closely. The strongest tooling runs on AWS, especially through Amazon Bedrock. That can speed deployment for AWS users. It can create difficult questions for operators using private clouds.
Ericsson says broader cloud-native deployment remains possible. Yet AWS appears to be the main reference path today. Carriers will need clear answers on portability and feature parity. They must know what changes when running outside AWS.
Overall, Ericsson is turning AI agents into telecom operating infrastructure. The plan links automation, customer experience, and revenue goals more tightly. If execution matches the blueprint, OSS/BSS may become far more dynamic. If not, operators may inherit another complex automation layer.
