Integrating genAI into customer support remains a key objective – and now there’s a simpler way to do it

Commercial context

As with many organizations within the broader tech sector, the dawn of generative AI (genAI) and machine learning (ML) has created a frenzy of expectation among CSPs about how these technologies can be incorporated into daily operations. A particular area of interest lies in using these technologies to anticipate customer needs accurately to help resolve any issues before they occur. Legacy methods fall short in providing valuable insights, hindering the ability to understand and meet customer expectations – but integrating genAI and ML may grant CSPs the power to foster stronger customer relationships and improve service quality.

The solution

The Beyond chatbots – Revolutionizing telecom with advanced generative AI Catalyst is providing the industry with a solution to this issue by using genAI and large language models (LLMs) to meticulously analyze customer care interactions, including call and chat logs. The core functionality involves accurately labelling keywords and phrases and associating them with topics, and reasons for the interaction and resolutions, all of which forms the basis for ML intent score models. These models facilitate a ‘virtual agent’ underpinned by genAI, which is capable of managing entire interactions. A standout feature of this approach is the 'intent and genAI factory,' a continuous cycle of identification, vetting, and development of additional intent and conversation models. This innovation ensures consistent enhancement of capabilities, creating the basis for viable long-term strategies for improving customer relationships.

This comprehensive solution is built on TM Forum assets, such as open APIs and Open Digital Architecture (ODA) best practices. The blueprint is designed for replication and adaptation across the industry, promoting collaborative innovation. The recommended use of three REST APIs—Get Customer Profile Data, Apply LLM, and Get ML Model Results—developed following TM Forum REST API standards, ensures consistency, interoperability, and ease of integration. Though not yet approved as ODA standards, these APIs offer substantial benefits, including enhanced data integration, improved accuracy in customer interactions, and the provision of personalized, proactive services.

Wider application and value

This Catalyst will significantly advance the cause of genAI and ML incorporation within the industry – particularly with respect to customer needs prediction and service quality enhancement. The solution demonstrated by the project team provides a foundation for the development of long-term implementations and contributes towards the broader goal of formalizing ODA standards. The business impact from early demonstrations appears substantial, particularly in customer service and call center operations. Notable measurable outcomes include a projected 40% reduction in call durations and a 30% reduction in customer return calls, leading to significant time savings and efficiency gains. Additionally, the solution is expected to improve the Net Promoter Score (NPS) by 4-5 points, indicating enhanced customer satisfaction and loyalty.

The financial implications are equally significant. As explained by Hatice Altınok, Senior Architect Intelligence Engineer at TurkNet, “by automating the diagnosis of internet connectivity issues and streamlining the troubleshooting process, the solution can lead to a 50% reduction in overall call center operational costs. These savings stem from decreased call durations, fewer return calls, and improved operational efficiency. Ultimately, the solution offers broad benefits to businesses, the wider industry, and society by enhancing efficiency, reducing costs, and elevating customer satisfaction. The Catalyst's innovative approach positions it as a critical driver of transformation in the telecommunications sector, and sets a new benchmark for customer service excellence and operational effectiveness.”