Feedzai has introduced Farol, an embedded AI agent designed to speed up bank fraud investigations and reduce manual workload, as financial institutions grapple with rising losses and compliance costs.
Feedzai has launched Farol, an embedded AI agent meant to help banks investigate fraud more quickly and ease pressure on risk teams as losses and compliance costs climb. The company said the tool works inside its RiskOps Studio platform, allowing analysts to stay in the same system while the software helps with case review, rule analysis and response drafting. According to Feedzai, the aim is to give fraud teams more output without forcing them to add headcount or switch between tools.
The release lands at a time when banks are facing heavier fraud-related burdens. Feedzai says reimbursement requirements are increasing institutions’ exposure to losses, while spending on technology and manual reviews continues to rise. The company also argues that many agentic AI projects have disappointed because third-party models do not have access to live transaction data, limiting their usefulness in real-world fraud operations.
Farol is designed to address that problem by sitting inside the bank’s own environment and working with the context available in existing workflows. Feedzai says its internal research found that 68% of financial institutions are trialling agentic AI, but that most of those efforts have yet to produce efficiency gains. The new agent is meant to bridge that gap with full audit trails, contextual analysis and the ability to act autonomously without requiring changes to core systems.
Feedzai says Farol includes several specialised functions. One reviews rule sets to identify noisy controls that generate alerts without catching fraud and suggests tighter thresholds, which the company says can reduce rule maintenance from days to minutes. Another gathers and summarises alert data for investigations, which Feedzai says has cut handling times by 20%. A knowledge feature answers product questions inside the interface, while a report-writing function can draft Suspicious Activity Reports up to 12 times faster by cutting down on manual collection and summarisation work. Chief product officer Pedro Barata said fraud teams need an agent that already understands their environment rather than another tool to manage, adding that the system is meant to surface useful information faster and reduce repetitive work.
Disclaimer: This article is intended to inform and educate, not to recommend or endorse any financial product, investment or strategy. Please consider your own financial circumstances and seek professional advice where appropriate before making financial decisions.
