An LLM-based AI agent to assist with smart deployments across business operations

Commercial context

As CSPs worldwide aim to integrate AI into their operations and business services, it's crucial that these deployments are strategically planned and adaptable to long-term challenges and varied use cases. Yet many AI deployments remain siloed within specific applications or suffer from limited efficiency – and this fragmentation creates difficulties in assessing return on investment (ROI) and hinders staff engagement with AI technologies.

This highlights the need for a more unified and strategic approach to AI adoption, which could serve as an industry-wide standard for integration. The emergence of large language models (LLMs) has introduced a new opportunity in the form of ‘AI agents’. An LLM-based agent can assist CSPs in integrating various AI deployments, even when facing highly specific, complex, and unique requirements. If such an agent were available industry-wide, CSPs could significantly enhance operational efficiency, improve customer experiences, and unlock new revenue streams.

How the solution works

Establishing a framework for such an AI agent has been the mission of the AI agent empowering higher autonomous level Catalyst. The project group has developed an advanced LLM-based technical architecture designed to simplify integration into the variety of CSP operational processes. This framework is underpinned by TM Forum’s Value Operations Framework, which includes its model and execution system, and other assets, including AIOps, DT4DI, and AOMM. These help ensure that the solution can be applied – and adapt – to various business scenarios.

Part of the project team’s main focus was creation of a finely-tuned LLM which can be incorporated with digital twins and domain-specific knowledge – something critical to enabling the AI agent to comprehend and operate effectively within the telco environment. These AI agents can be applied across multiple value streams within the telecom operational cycle and customer experience lifecycle, enhancing everything from provisioning and installation to after-sales service and incident management.

Wider application and value

The implementation of this LLM-based AI solution holds considerable promise for CSPs across the globe. For instance, China Telecom plans to use the agent to significantly enhance its AI and cloud network operations. It’s also expected to support 35 new beyond-connectivity products for over 190 million wireline broadband subscribers, assist with the automation of 75% of cloud network incident handling, and improve the efficiency of high-risk command audits by 50%.

Moreover, China Telecom expects it will reduce hardware-related on-site visits for home broadband installations and maintenance by 50%. The Chinese CSP also expects an increase in the number of base stations maintained per capita by 21%, and a cut in nationwide operations and maintenance (O&M) costs by 200 million RMB annually. There are considerable environmental benefits too, with the CSP claiming that this will help achieve an annual energy saving of 760 million kWh across 5 million base stations and 90 million kWh across 2,500 IDC facilities, while boosting the autonomous operation level to L4+ in certain cloud network scenarios.

Similarly, HKT's use of the agent for VVIP user experience assurance has yielded impressive results, with issue handling efficiency improved by 85%, reduced processing times from hours to minutes, and halving the bad customer experience indicator from 2.7% to 1.3%. Overall fault resolution efficiency has seen a 30% improvement, leading to annual O&M cost savings of 15 million RMB and achieving higher customer satisfaction ratings than competitors.

According to Project Leader, China Telecom’s Zhang Le, “the broader implications of this technology are vast, and provide CSPs with a strong framework to innovate and transform their operations. By harnessing the power of LLM-based AI agents, CSPs can achieve significant cost savings, enhance operational efficiency, and deliver superior customer experiences, all the while contributing to sustainability goals through substantial energy savings too.”