Your AI Journey: Finding the Use Cases That Matter
Improving operational efficiency with AI is non-negotiable in telecommunications. Discover how targeted AI use cases can deliver immediate value.
Blog
23 February 2026
Author
AI has rapidly gone from buzzword to ubiquity. As 2026 approaches, the real question is: where is the true value, and how do we unlock it?
Blog
23 February, 2026
Author
AI Isn’t Just a Buzzword. It’s a Tool that Should Generate Returns.
Artificial Intelligence has long been a buzzword, but it has also now become ubiquitous. Telcos have made and are seeking to make significant investments in AI – so the expectation is set: AI must generate returns in-line with other investments. With that focus in mind, where is the value and how can we unlock this in 2026?
AI isn’t just a tool. It’s a fundamental element of transformation that our industry must embrace. To do so, we must collaborate and share lessons, so we can, collectively, make progress and turn vision into practical reality. In the spirit of this, in this article, we’d like to share some key learnings from our own AI journey that can drive success in the year to come.
AI Adoption Needs to Scale to Ensure Returns
We all know that AI and ML will become central to telecoms networks. There’s very little dissent on that now – but there is still plenty of room for debate regarding the pace of change.
That’s because the real questions are:
-
-
Can we demonstrate tangible, repeatable value from the capabilities it brings, and
-
When will this value be captured sufficiently to make a difference?
-
This uncertainty is real – despite recent gains. Foundation models have vastly expanded their parameter scales, their multimodal competencies and context windows – by orders of magnitude. This rate of improvement is without precedent in the history of software systems – and enhancements across multiple dimensions are likely to be seen in 2026.
The results of this can be seen in rapidly advancing industry benchmarks and also the increasingly sophisticated behavior of modern large language models. It’s also clear from the growing number of use cases – AI evidently has a role to play almost everywhere.
At Polystar, we already expect results from operational use cases.
For example, AI tools can help us to accelerate the identification of emerging network and service issues, enabling us to summarize complex network phenomena across multiple domains, and resolve incidents with new levels of speed and reliability.
Equally, many operators have invested in AI to support customer-facing applications and service touchpoints. These investments will scale from trial cases to reach widespread adoption. For example, innovations — such as the ability to improve NPS through accurate sentiment analysis at true scale, AI enhanced digital interactions and customer journeys, and more intuitive and responsive sales and support experiences — are already showing how AI can deliver value quickly.
All of this means that operators are gaining clarity over how AI-enablement can benefit key processes and operations.
The next step is crucial: converting promising PoCs into deployments that deliver positive operational impact.
User Interfaces and User Experience – Tightly Linked
A good deal of recent activity regarding AI in telecoms has been focused on information retrieval, using text (or voice) based chat interfaces to allow humans to interact with AI-enabled systems. This considerably simplifies the way in which information can be gathered – and delivers consistent responses.
But telecom networks and operations need more. Our operational domain does not simply include text-based information or interrogation. We have to deal with information from a wide range of sources - packets, telemetry, performance metrics, configuration and location data.
Visualize Complex Logical and Physical Relationships
Telecoms operational information can also be described as being inherently visual in nature. It includes topologies, alarms, service paths and flows, node and domain relationships and hierarchies, geolocations, and more — all connected through complex logical and physical relationships. These patterns are best understood through interfaces that visualize them in domain-specific ways, rather than through text alone.
So, we think that a key opportunity for productivity gains is to combine domain-specific UIs with AI models that can interpret and act on the specialized data collected and presented. This will apply to both ad hoc interpretations (for example, a user might ask “show me what’s degrading and where”) and also emerging agentic AI workflows (“fix this under my supervision”).
As networks move towards machine-driven operations, this new industry- and domain-specific kind of interface will become essential for trust, clarity, and control. These new interaction models will also redefine how engineers work.
Identifying and Tagging Data
Combining and correlating customer-experience outcomes with network and operational data will be critical for maximizing efficiency. To achieve this, we need to link customer-facing experience indicators - such as NPS scores, customer-care interactions, and sentiment - with network performance metrics like KPIs, alarms, logs, and drive-test results.
ML/AI will help us to align these perspectives because it offers new levels of processing capabilities. When we can do this, we’ll be able to build new models that show how, for example, NPS might be enhanced with specific KPIs. Ultimately, this will help us to allocate capital more efficiently, to drive and sustain competitive advantage, and to deliver meaningful OPEX reductions.
Define our Goals to Secure the Right Outcomes
Competition continues to be intense. Not only from peer operators, but also from emerging Non-terrestrial Networks that offer satellite to cell connectivity, digital MVNOs, private network providers, and more. If we also consider that, to date, 5G investments have seen relatively modest revenue gains, we can see that the pressure our industry faces has grown significantly.
So, while CAPEX is tied to regular cycles of investment in new generations of technology, as well as regulatory drivers for coverage and service performance, OPEX reduction has become a key focus. Can future operations be delivered at a fraction of today’s cost base, given the ongoing need for investment in tech evolution?
We need to learn if emerging operators — for example, satellite-to-cell providers that have made significant investments in their orbital networks - can achieve more profitable OPEX ratios.
Will they challenge classical MNOs and gain market share? Ambitious efficiency goals could well be the catalyst that drives innovation and secures the desired outcomes.
