For Harsha Angeri, VP, corporate strategy & head of AI business at Subex, the solution lies in one of the most transformative technologies of our time: Generative AI.
“Fraud is no longer simple — it’s multifaceted, rapidly evolving, and difficult to catch using traditional tools,” Angeri tells Capacity.
“Most CSPs [communications service providers] also struggle with a lack of data science talent. That’s where GenAI steps in — not just to fill the talent gap, but to transform how we even approach fraud detection.”
Subex’s GenAI-based agent squads monitor diverse data sources 24/7, fusing signalling, usage patterns, and customer behaviour to spot anomalies in real time. The agents employ autonomous learning, enabling them to evolve alongside emerging fraud tactics — without the need for manual rules or static thresholds.
“These agents are continuously learning and adapting. Their ability to reason using large language models (LLMs) gives CSPs a dynamic defence against increasingly sophisticated scams,” Angeri explains. “What’s more, we use natural language insights to make it easier for fraud cases to be understood and investigated by non-technical teams.”
From tribal knowledge to scalable intelligence
Subex’s AI-led approach is not just about fraud prevention — it’s fundamentally changing how telecom operators manage risk and assurance.
“One of the biggest shifts we’re enabling is moving from tribal knowledge to system-driven intelligence,” Angeri says. “We’re codifying expert heuristics into AI logic that can scale, bringing consistency to fraud detection across teams, geographies, and time zones.”
This shift also enables faster onboarding of fraud analysts. With guided recommendations and real-time AI support, manual reviews are being replaced with intelligent, automated responses — a crucial capability in a high-stakes environment.
Future-proofing through openness and cloud agility
Another pillar of Subex’s strategy is ensuring its solutions remain future-proof through open ecosystem integration and multi-cloud readiness.
“We’ve adopted CAMARA APIs for key functions like SIM swap detection and location awareness,” says Angeri.
“While the ecosystem is still maturing, we’re actively working with it and also integrating additional third-party data sources to enrich our detection capabilities.”
Subex’s flagship platform, Hypersense, is cloud-native, enabling seamless deployment across multiple environments. This ensures flexibility for CSPs as their infrastructure and digital transformation journeys evolve.
Beyond fraud, GenAI is being applied to transform the traditionally reactive domain of business assurance into one of strategic innovation.
“Our core proposition is to unlock assurance data for multiple use cases — not just for compliance, but to drive margin and revenue optimisation,” Angeri explains. “Context-aware anomaly detection now replaces outdated static thresholds, while GenAI can autonomously write controls when deviations occur.”
Other innovations include AI-generated investigation playbooks tailored to each anomaly, as well as AI copilots that drastically reduce investigation times and enhance audit trails.
GenAI also enables direct action — from sending alert emails to logging tickets — boosting responsiveness and reducing operational drag.
Why data strategy is the bedrock of AI success
Despite these advances, Angeri is clear that no AI deployment can succeed without a robust data foundation.
“CSP data is incredibly fragmented — it lives in legacy systems and isn’t standardised. Without clean, connected, and contextual data, GenAI won’t deliver value,” he warns.
To solve this, Subex’s platform unifies data from siloed sources using telecom-specific ontologies, while also providing observability tools and data contracts to ensure data integrity. Domain-aware data pipelines further streamline the preparation of telecom features for LLMs.
“Ultimately, GenAI is only as good as the data it learns from,” Angeri concludes. “That’s why we focus not just on building AI, but on making the data ecosystem robust enough to support it at scale.”
RELATED STORIES
Fostering innovation in tech: An Interview with Nisha Dutt
Proximus Global launches AI-powered anti-fraud platform to combat SMS scams
