BARCELONA, Spain, May 29, 2025 -- Operating telecommunications networks is getting more difficult. Networks are converging: fixed and mobile are often both required to provide a given service experience, whether it's an IoT use case or end-to-end network slicing.

As networks become more software-driven, operating them requires different skills in maintenance, and assurance. Add to that complexity new technologies like enterprise-oriented APIs and agentic AI. These technologies can optimize network operations for granular service differentiation and prioritize business KPIs over simple uptime. With the right architecture and operational practices, networks can serve much more diverse use cases with a higher degree of reliability.

Huawei Can Help

Huawei Intelligent Operations works closely with its own network technology engineers as well as with industry standards bodies to guide the evolution of network services, and to ensure it supports all industry best practices. It works closely with the TM Forum, coordinating customer inputs and its own operational vision to create service-centric operational metrics. Thus, Huawei is expanding the boundaries of AI in telecoms operations, devoting intensive research into digital twins and telecoms-specific foundation models. Additionally, it feeds its substantial experience running telecoms networks back into its own operational best practices.

Key examples:

In the Asia-Pacific region, Huawei collaborated with a carrier to analyze the service impact of faulty sites. The work made use of Huawei's Expected Demand Not Served algorithm, which identifies service usage that is lower than usual, identifies faults that may be leading to depressed usage, and prioritizes fixes to return the network to its designed functionality. This has reduced traffic loss by 15% for this carrier.

In the Middle East, Huawei Intelligent Operations significantly improved topology accuracy for an FTTx network, using delay-tolerant network architecture to reduce invalid work orders by 60%, and ultimately, led to a 10% decline in customer complaints.

In Europe, Huawei used Gen-AI to improve MTTR by 25%. Huawei's work with the carrier on role-based copilots for field maintenance personnel and multi-agent collaboration to provide comprehensive treatment of complex scenarios.

Kevin Ye, President, Intelligent Operation Domain, Huawei

"Huawei Intelligent Operations believes not only in helping its CSP partners conquer its existing operational challenges, but in helping them harness the latest technology to drive new business benefits across the company" – Kevin Ye, President, Intelligent Operation Domain, Huawei

Gen-AI Applications and Challenges

The dramatic rise of generative AI has led to two basic operational models: copilots and agents. Copilots help a human employee for coding, configuring systems, or managing a stream of service quality data and trouble tickets. Agents are more devoted to a specific set of tasks or operational concerns, and can operate independently under the supervision of a human employee.

In order to use Gen-AI, these tools must address the following concerns:

  • Hallucinations. Gen-AI can generally produce wrong information. Therefore, model training, retrieval-augmented generation, verification from other systems, and other guardrails are all required to minimize incorrect data and actions.



  • High cost. Gen-AI can become expensive. CSPs need an AI version of finops that minimizes the cost while still reaping the benefits of this new technology. For example, services partners can determine the lowest version of an LLM that still produces the desired output, and systems can be designed to minimize token usage.



  • Integration with existing systems. Agentic AI is already demonstrating that a Gen-AI interface couple with a digital twin of the network can produce powerful automation. Integrating data sources and BSS/OSS systems is complex work, however.



  • Data transformation and management. Building that digital twin requires breaking down data siloes, cleaning and normalizing bad data, and bringing together disparate data formats and sources in a single repository. This requires experience and strong professional services.



  • Training. Without training from a domain expert, LLMs can't understand CSP's documentation, trouble tickets, operational processes etc.

Huawei Intelligent Operations can help.

DeepSeek: The Next Generation of Gen-AI

The January 2025 introduction of DeepSeek proved that Gen-AI could provide all of its current capabilities with a much more efficient model. Huawei's research indicates that DeepSeek provides advantages such as domain knowledge construction and application, code generation, data analysis, and intelligent ticket processing have direct application to CSP operations. Huawei is at the forefront of applying these new models to telecoms infrastructure.

Huawei's Vision for ICT Operations

Huawei believes that ICT technology will continue to play an important role in driving the digital economy, with new technologies acting as intelligent accelerators. Reliable network operations and intelligent operations will continue to grow in importance. Huawei continues to invest heavily in R&D to support these advancements, and will continue to accelerate the transformation toward intelligent operation.

Recently, Huawei has collaborated with TM Forum and leading carriers to release annual versions of the New-Generation Intelligent Operations White Paper 3.0. which provides valuable insights into the latest business practices and applications of new technologies.

Photo - https://mma.prnewswire.com/media/2698625/Kevin_Ye_President_Intelligent_Operation_Domain_Huawei.jpg

Cision View original content:https://www.prnewswire.co.uk/news-releases/huawei-intelligent-operations-advances-ai-and-gen-ai-in-ict-302468043.html