Local Startup Leverages AI Agents To Enhance Fraud Prevention in Qatar Payment Systems

Doha-based fintech Dibsy is developing an AI agent-powered prevention technology purpose-built for the high-velocity world of modern payments, delivering real-time protection while ensuring a smooth customer experience, its chief executive has announced.

In a statement to Gulf Times, Dibsy CEO Loyan Farah emphasised that in today’s hyper-connected payment ecosystem, fraudsters are not acting alone. “They are organised, well-funded networks leveraging sophisticated tools and exploiting vulnerabilities in milliseconds. For payment providers in Qatar, the challenge is not just keeping up — it’s staying ahead,” Farah pointed out.

According to Farah, traditional rule-based fraud prevention systems, built on static thresholds, rigid policies, and periodic manual updates, “are no longer sufficient.”

“As fraud tactics evolve daily, we need a new approach that combines speed, contextual intelligence, and continuous learning.

 The answer is Agentic AI: specialised, autonomous AI agents designed to actively monitor, analyse, and respond to fraud threats in real-time,” he explained.

Farah further explained why traditional fraud systems fall short, hindering their effectiveness in today’s fast-paced digital environment. One major limitation is their reliance on static rule dependencies, which cannot adapt quickly to emerging fraud patterns, he noted.

“Additionally, the delay in reaction times poses a problem considering that modern payment processes require instant decision-making, while manual reviews typically struggle to keep up,” Farah emphasised.

Because traditional fraud systems often operate in isolated data silos, Farah noted that transaction data, behavioural signals, device fingerprints, and identity checks often operate in disconnected systems.

He also underscored the issue of high false positives, saying “overly strict rules” lead to unnecessary declines of legitimate transactions, resulting in frustrated customers and lost business opportunities.

“Old systems were designed for yesterday’s fraud. But today’s fraudsters leverage machine speed, global networks, and highly adaptive tactics. Static defences simply break,” Farah stressed.

According to Farah, AI agents bring a new level of capability to fraud prevention. Unlike traditional AI models that wait for instructions, he said these agents operate with a degree of autonomy, constantly scanning for threats, analysing multi-dimensional data, and adjusting responses in real-time.

“At Dibsy, our multi-agent fraud prevention system includes specialised AI agents designed for different fraud types. The first is our Transaction Fraud Agent, which continuously monitors every payment in real time and detects abnormal transaction patterns, such as sudden spikes in value, unusual payment routes, or velocity anomalies. It also dynamically adjusts risk scores as new signals emerge.

“The second is Dibsy’s Behavioural Biometrics Agent, which establishes baselines for individual user behaviour — from typing speed to app navigation. This agent also identifies anomalies that may indicate bots, mules, or scripted fraud attempts,” Farah further explained.

Farah also underpinned the importance of human oversight in deploying AI agents, citing “augmented intelligence,” rather than “full autonomy”, to ensure that these agents do not operate in isolation while automating various tasks.

He pointed out that Dibsy’s approach is characterised by “full transparency,” where every decision made by the AI agent includes detailed logs that state the reasons behind triggering alerts.

Additionally, Farah emphasised that these agents are designed for “continuous learning,” allowing them to refine their models over time based on actual outcomes, thus improving accuracy without requiring full retraining cycles.

Furthermore, “controlled escalation” ensures that only complex or ambiguous cases are elevated to human analysts, allowing for efficient collaboration between AI and human expertise.

Farah added: “We believe in augmented intelligence, not blind automation. Machines handle scale and speed, but final judgment always stays with humans.”