How Deutsche Telekom is Rolling Out AI Agent for Network Ops

Deutsche Telekom activates its AI-powered RAN (Radio Access Network) Guardian Agent to support mobile network operations across Germany. 

The agent monitors mobile network activity, detects irregularities and takes corrective actions in real time, reducing delays in fault resolution and improving service continuity during large events or unexpected spikes in demand.

AI takes a bigger role in network management

Unveiled at Mobile World Congress alongside Google Cloud, the RAN Guardian Agent is the first of its kind to be introduced by a telecoms operator.

The tool automates core functions in network maintenance using a multi-agent system that processes public data, monitors mobile traffic and executes optimisation tasks. This is seen as a first step toward creating autonomous, self-healing networks.

Abdu Mudesir, Board Member for Product and Technology at Deutsche Telekom, says: “With the introduction of the RAN Guardian Agent, we are the first network operator to rely on a highly developed AI agent in network management. 

“Our development teams have thus done pioneering work in the development of AI agents for mobile networks. With an intelligent interaction between our network experts and AI, we are solving specific challenges for the benefit of our customers – for the best network. And we are taking a big step towards autonomous, self-healing networks.”

Layered AI agents drive network response

The RAN Guardian Agent uses a layered process to predict and manage network usage. In the first step, the AI scans open data sources, including online directories and social media, to identify upcoming events across Germany. It compiles a verified list, estimates potential attendance and determines precise locations.

Next, the platform evaluates how local mobile networks are likely to perform during these events. The AI checks antenna capacity and other key parameters, detecting potential congestion or service degradation before it happens.

In the final step, a dedicated agent initiates optimisations. These include reallocating network resources or modifying existing configurations. 

Each action is documented to feed performance data back into the system, allowing continuous refinement and learning.

This system design makes it possible for Deutsche Telekom to handle high volumes of mobile traffic without relying entirely on manual interventions. 

By combining autonomous monitoring with automated decision-making, the agent strengthens resilience in network operation, especially in scenarios of sudden load change or widespread user activity.

Improving efficiency across telecom operations

Telecom networks have grown in complexity and existing automation strategies can no longer support all real-time management tasks. The RAN Guardian Agent uses large language models and advanced reasoning tools to extend the capabilities of standard automation by introducing systems that interpret data, make decisions, act independently and validate each other’s responses.

This new method reduces human intervention, improves time to resolution and maintains service levels even in challenging conditions. 

Tasks that once required an hour of human-led analysis can now be completed in minutes, supporting Deutsche Telekom’s goal of delivering a more consistent and scalable mobile experience.

Through AI integration, the operator can anticipate issues instead of reacting to faults after service degradation has occurred. It contributes to both operational efficiency and customer satisfaction, with optimised performance delivered where and when it is needed most.

Technical foundations of the RAN Guardian Agent

The RAN Guardian Agent is built using Google’s Gemini 2.5 and is supported by a suite of Google Cloud services including CloudRun, BigQuery and Firestore. These components enable the system’s real-time performance and decision-making capability.

Key technical functions include autonomous RAN monitoring, AI-powered classification of network issues and proactive optimisation through autonomous configuration.

Data is sourced from multiple layers, including live performance and coverage reports, network inventory and monitoring tools.

This integration of cloud‑based platforms with telecoms operations marks a step forward in applying AI for live mobile environments. 

It opens new possibilities for telcos to manage expanding traffic, edge deployments and high‑density networks with less manual input and faster fault correction.