Agentic AI is only as reliable as its data

Artificial intelligence (AI) is driving a transformative shift in the telecommunications industry. While generative AI has been primarily used to boost operational efficiency, agentic AI is emerging as a breakthrough technology, slated to redefine how telecom providers operate. For instance, it is projected that agentic AI will manage 68% of customer service and support interactions, resolving some 80% of common service issues due to its autonomous nature.

What is agentic AI?

Agentic AI pertains to artificial intelligence systems designed to operate autonomously, making decisions and performing actions with minimal human intervention. Agentic AI combines multiple AI models capable of interpreting data, executing tasks and reacting autonomously. Unlike traditional systems that merely recommend actions, agentic AI actively plans, decides, and executes the best course of action, adapting in real time based on changing data and outcomes. These agents gather data from diverse sources and operate in real-time with minimal human input. They can also interact with one another to manage complex, multi-step workflows, uncover trends and patterns that traditional tools may miss. In the context of telecoms, telco networks have the potential to metamorphosize into truly autonomous networks, offering systems that can handle intricate operations, recognize priorities, and proactively adjust themselves as needed.

Data is critical

Agentic AI relies on data for its continuous, dynamic decision-making. One advantage operators have is they are sitting on vast amounts of data, exacerbated by the millions of user connections and touchpoints. This is, however, expected to increase exponentially, with global data volumes projected to surpass 5,4000 exabytes by 2030, driven by the rapid development of AI-powered technologies, smart devices and IoT systems. A challenge for telecom operators is how best to ingest and leverage these enormous quantities of data for agentic AI. Operators need to not only store and process, but they need to collate and ensure the quality of the data stored in silos and spread across the network. This is a complex and time-consuming process.

Avoiding the bias

An additional conundrum telecom operators face once they’ve ingested large data sets for agentic AI, is how to ensure the agents don’t replicate or exacerbate biases? Like with generative AI, addressing and avoiding biasness is an essential and often complex task. AI agents develop emergent behavior and if the data they process is unreliable, the agents will learn, amplify and reinforce systemic biases over time when left to optimize complex systems. This emergent unintentional bias, driven by data imbalance, can compromise network quality and exacerbate service disparities.

Realizing agentic AI’s full potential

Realizing the full potential of agentic AI means telecoms need a foundation layer built on deep, end-to-end, real-time insights into the network operations, with a comprehensive understanding of the customer experience. This means granular, subscriber-level data that can be analyzed and correlated with network performance to identify issues such as why customers are likely to churn and what actions need to be implemented to strengthen customer loyalty.

The data analyzed needs to include everything such as multiple KQIs and KPIs, device types, days of the week or time of day from point-to-point across the entire network. The deep insights offered will then be able to power agentic AI’s reliable reasoning and automation capabilities. With trusted, correlated data, agentic AI can fuel everything from proactive customer care to behavioral prediction and precise root-cause analysis, potentially resolving uses before a customer even reaches out.

Assurance solutions: the gold-data standard

Intelligent assurance solutions incorporate real-time user and service analytics and embed a data and insight foundation layer to enhance operators’ network and customer experience. The end-to-end solutions are already designed to monitor tens of millions of connected devices, high-speed voice and data services and changing geographical conditions. These service assurance solutions shine a light on both the network and the customer and make use of insights into user trends, location intelligence, usage patterns, service interactions and more.

Embedding an agentic AI layer on top of this enables operators to utilize the existing data on their networks, ensuring reliable and trustworthy insights from real-time balanced data models. These agents can then identify real trends, locate and predict potential faults and provide premium services for customers. The adeptness of agentic AI to collect data from different sources also means that assurance solutions can be integrated with other management systems, prioritizing the subscriber experience.

In conclusion

Agentic AI is truly disruptive. It is expected to lead to a 30% reduction in operational costs and greatly improve customer satisfaction, which is at an all-time low across all businesses and industries, not only telecom. However, the output of agentic AI can only be as credible and trustworthy as its inputs, i.e. the data that it is fed. Assurance solutions have been facilitating the seamless flow, processing and interpretation of real-time, correlated and end-to-end data and insights for quite some time already. Combining agentic AI will truly transform the network into efficient and autonomous systems.