Vendors must adopt GenAI solutions quickly to maintain a competitive edge in the telecoms market

Generative AI (GenAI) is a game-changer for B/OSS vendors, offering a critical path to staying competitive in the telecoms market. These vendors, providing essential business and operations support systems to communications service providers (CSPs), are now using GenAI to ensure they remain relevant in the fast-paced world of telecoms applications.

To delve deeper into the role of GenAI in the telecoms applications market, we conducted a comprehensive survey of ten leading B/OSS vendors. This unique survey, conducted in 2Q 2024, explored vendors’ GenAI strategies, motivations and challenges. Key questions addressed include: what are the primary drivers for adopting GenAI technologies? What are the key benefits and challenges associated with using GenAI in B/OSS systems? What development strategies are vendors adopting to use GenAI effectively? And what are the most promising use cases for GenAI in telecoms applications, and how are vendors monetising these opportunities?

The findings, detailed in our full report, offer valuable insights into the industry’s current state and future trends.

Improving customer experience is the key driver for vendors’ GenAI activities

Improving customer experience emerged as a key driver for vendors’ GenAI investments, with a focus on enhancing both customer engagement and core product functionalities. According to our survey, 70% of respondents identified GenAI’s ability to significantly improve customer interactions with existing offerings as a key driver for their investment (Figure 1). This aligns with the 70% of respondents who prioritised enhancing key functions within their current products. By using GenAI’s natural language capabilities, vendors aim to address customer challenges related to technical expertise, empowering non-technical staff to perform complex tasks more effectively. This ultimately translates into a more intuitive and personalised customer experience. All respondents indicated involvement with GenAI,¹ suggesting a high level of interest in the technology. While all vendors that responded to our survey reported involvement with GenAI, it is important to note that some may be doing so primarily to align with current AI trends or to avoid the perception of inaction. 

Figure 1: Key data points related to telecoms application vendors’ use of GenAI 

While drivers for investment are clear, vendors did not identify a singular, outstanding challenge hindering the development of GenAI-capable applications; instead, their concerns were more evenly distributed. Legal and regulatory issues remain a key challenge for many vendors, with some believing that CSPs should take the lead in addressing these issues. Vendors highlighted other challenges, such as limited data access and data privacy concerns. While vendors have overcome initial hurdles related to model selection and skill shortages, the evolving regulatory landscape and the complexity of GenAI applications continue to present significant obstacles.

Telecoms application vendors use customised models to develop GenAI use cases

The vast majority of telecoms application vendors (80%) rely on customised foundation models (FMs) for their GenAI applications (Figure 1). This preference for customised models is primarily driven by the prohibitive cost of building proprietary FMs. To meet their specific use case requirements, 90% of respondents collaborate with multiple FM providers and tailor these models accordingly. This approach allows vendors to use the strengths of different FMs while ensuring that the models align with their unique business needs. 

Vendors predominantly customise FMs by modifying prompts, employing either prompt engineering or prompt learning techniques. This approach offers a distinct advantage, enabling a multi-FM strategy. By focusing on prompt modification, vendors can address a wider range of GenAI use cases without the complexities involved in fine-tuning model parameters. This flexibility allows them to use the strengths of different FMs while maintaining control over the model’s behaviour. Model customisation, primarily through prompt modification, offers several significant benefits. By tailoring FMs to specific use cases, vendors can save valuable time and reduce the likelihood of hallucinations or unintended outputs. Additionally, customisation helps mitigate data safety concerns by ensuring that the model’s behaviour aligns with the vendor’s privacy and security requirements. However, it is essential to implement robust safeguards to prevent prompt modification from leading to the generation of harmful, biased or personally identifiable information.

Customer engagement and service orchestration use cases are key to telecoms application vendors’ GenAI strategies

Telecoms application vendors are using GenAI for applications that improve customer experience. This is especially true in areas like customer care and customer engagement. Thanks to advancements in natural language processing, these tools can have more natural conversations with these teams and provide better information. Beyond direct interactions, GenAI is also being used to streamline back-end processes, with vendors deploying it in service orchestration. GenAI can understand complex customer requests and suggest the right services and pricing. It can also help manage contracts by making changes and finding contract terms or can generate quotes and proposals more efficiently.

While vendors have made significant strides in implementing GenAI for improving customer engagement and streamlining service orchestration, the survey results reveal a noticeable gap in the development of virtual assistants that directly interact with customers. This lag can be attributed to several factors. First, the surveyed entities primarily specialise in telecoms applications rather than virtual assistant development. Second, CSPs may be hesitant to expose customers to GenAI-powered virtual assistants until the technology matures and its capabilities are fully demonstrated. The history of less-than-stellar performance with previous virtual assistant implementations may have made CSPs cautious about rushing into adopting GenAI-powered solutions for the same use case. 

¹ Involvement with AI includes developing products, looking at partnerships, understanding the AI ecosystem, etc.