Making the most of GenAI in telco and cableMaking the most of GenAI in telco and cable

There is probably no hotter topic in the telecom industry today than generative AI (GenAI). And for good reason: GenAI has the potential to boost employee productivity, operational efficiency and customer service throughout the entire telecom sector.

More specifically, GenAI has the power to transform the telecom business by automating more business processes and analyzing customer data much more efficiently than ever before. The technology should enable providers to correlate information about their customers and operations better and faster, thereby enabling them to communicate much more effectively with subscribers.

To back up their case, GenAI proponents point to studies indicating that the use of the technology leads to much greater business and worker productivity, fewer network performance problems and higher customer engagement. In the latest survey of service providers conducted by Omdia, for instance, respondents cited top productivity applications such as automated code development, chatbots, virtual assistants and content generation, among others.

Thanks to these benefits, Gen AI promises to be of enormous importance to telecom providers. That explains why leading telcos around the world, such as BT, Vodafone, T-Mobile, China Telecom, du and others, are embracing the technology so enthusiastically. As a result, Omdia projects that global GenAI software revenue will jump at a 53% CAGR from 2023 to reach $58bn globally by 2028.

Still barriers to overcome

Leveraging GenAI effectively, however, is no easy matter. Telecom providers must wade carefully into this space. While GenAI large language models (LLMs) have reached maturity, they are still far from being a ready-to-use service in telecoms.

And several major barriers must be overcome before telcos can take advantage of GenAI's many benefits. First, providers must determine their LLM strategy. They need to decide whether to build their own in-house model or leverage one or more of the public models available from third parties, such as Microsoft with OpenAI, Amazon Web Services with Bedrock, Google with Gemini and other major players. A keen grasp of the types of use cases needed, an understanding of which models can to the job best and measurements of the likely ROI will be critical for moving forward with LLMs.

GenAI technology is developing at a fast pace, and many different stakeholders are involved. These factors, as well as the lack of in-house skills and the huge cost of developing the technology internally (an estimated $200,000 a day), mean that most providers are likely to choose one or more of the publicly available LLMs and adapt those models for their own use.

Next, telcos must grapple with the increasingly troublesome issues of consumer privacy and data security. They must figure out how to keep their sensitive customer data safe and secure from harm and theft. This issue is particularly paramount with public LLMs, which will have a greater risk of data leaks or outright theft.

Third, providers must determine how to ensure the accuracy of the responses they receive to their GenAI queries. They need to make sure that the responses to queries are always correct and that the number of interactions required to obtain a correct result is not more expensive than the manual, human-centric methods used today.

Perhaps most importantly, telecom providers need GenAI solutions that will glean telco-specific data to be effective. Because publicly available LLMs come with no specific telco knowledge, they must be supplemented with more specialized data and tools (such as prompt engineering and retrieval augmented generation) to make them work in the telecom space.

This means accessing valuable and highly sensitive telco data from business support/operations support systems (BSS/OSS). In fact, Netcracker has found that a whopping 90% of telco use cases across all areas of the business need BSS/OSS data — often in real-time. However, telcos must work to avoid GenAI and data silos across their operations. That is because industry vendors tend to have data silos tied to their own BSS/OSS.

Common platform approach needed

Building and running a GenAI assistant for one use case and then employing another platform for the next assistant and use case will not yield successful results and is unlikely to generate a positive ROI. Providers will need a common platform approach for GenAI that works across their entire multi-vendor business.

In other words, telcos need help harnessing the awesome power of GenAI because they cannot do so easily on their own. Fortunately, GenAI solutions specifically adapted for telecoms with ready-to-go use cases are available from vendors like Netcracker. These solutions are finally taking GenAI out of the lab and into the real world of telecoms.

This blog is sponsored by Netcracker.