What challenges did Ericsson face before AI implementation?

Maintain optimal RAN performance and reduce complexity of real-time network optimization and troubleshooting as networks scale. Reduce time spent on analysis and decision-making while improving operational efficiency.

What was the impact of the AI solution?

Ericsson reports an 80% reduction in time spent on analysis and decision-making. The approach is positioned to improve OPEX efficiency, development/deployment speed, scalability, and network performance.

Rank#436of 3,449

Source

ProductionEvidence: Medium50/100

Ericsson agentic AI for autonomous RAN optimization using Amazon Bedrock Agents

Use case typeAutomotive operations automationUpdated Jun 13, 2026

Ericsson describes an agentic AI architecture for Radio Access Network optimization that uses AWS infrastructure and Amazon Bedrock Agents to coordinate specialized agents. The system includes a supervisor agent, cell anomaly detection, root-cause explanation, general and specialized optimization, and a natural-language 'Talk to Network' interface supported by RAG and visualization/reporting.

Organization: Ericsson
Industry: Tech & Comms
Location: Sweden
Published: July 2025

Reported outcomes

−80%

timeTime & speed

Strategic outcomes

Speed & agilityReduced analysis and decision-making timeNew product / capabilityAutonomous RAN optimization architectureNew product / capabilityNatural-language network interfaceScale & capacityClosed-loop autonomous network optimization

Catalog median for time & speed deployments: −50% across 312 reported metrics. Compare benchmarks →

Why do we believe this?Outcome claims, sources, and evidence checks

Normalized claim

Time: 80% decrease

Ericsson BlogJul 29, 2025Blog postInferred claimMedium evidence strength

Ericsson reports an 80% reduction in time spent on analysis and decision-making.

Why do we believe this deployment?Customer identity, provider attribution, maturity, and source checks

Customer: Ericsson
Provider: AWS
Maturity: Production
Linked source: Ericsson Blog

Reduce time spent on analysis and decision-making while improving operational efficiency

Customer identity supportedSource describes one deploymentMaturity supported

Primary read

Use case focus

Showing 3 of 3

1Agentic AI
2Network Operations
3Telecommunications

Maintain optimal RAN performance and reduce complexity of real-time network optimization and troubleshooting as networks scale.
Reduce time spent on analysis and decision-making while improving operational efficiency.

Implement a GenAI-powered supervisor agent orchestrating specialized agents for anomaly detection, root-cause explanation, and optimization.
Use Amazon Bedrock Agents and AWS foundation models with Talk2Data, RAG components, and reporting/visualization outputs.
Apply the architecture to closed-loop, intent-based optimization toward TM Forum autonomous network level 5.

Technologies

Amazon Bedrock Agents AWS foundation models RAG Talk2Data OpenSearch VectorDB Amazon Bedrock Knowledge Bases

Ericsson reports an 80% reduction in time spent on analysis and decision-making.
The approach is positioned to improve OPEX efficiency, development/deployment speed, scalability, and network performance.

Architecture

A hierarchical agentic AI architecture where an AWS Bedrock-based supervisor agent orchestrates specialized agents for cell anomaly detection, root-cause explanation, general optimization, and outputs. The solution uses Talk2Data and RAG connectors to rApp documentation/data, with OpenSearch/VectorDB, Amazon Bedrock knowledge bases, and reporting/visualization functions to support autonomous RAN optimization and intent-based closed-loop actions.

Sources & evidence1

Evidence: Medium50/100Evidence strength