Feedzai launches Farol, an AI agent platform for fraud teams, at Fusion London

Farol brings agents into fraud operations

Feedzai introduced Farol through an immersive video experience on the final stage of the event, held at Frameless in London on September 23. Bizarro guided the audience into what the company calls the agentic era. Farol means "lighthouse" in Portuguese.

According to Feedzai, Farol puts AI agents to work in fraud operations. They help teams research risk, connect investigations and reporting, and manage rulesets conversationally. The company says the aim is to focus human expertise where it matters most: less time searching and maintaining, more time on strategy, judgment and emerging threats.

Bizarro framed the agentic era around three goals:

• Provide agentic capabilities that help financial crime teams work smarter and faster.
• Enable teams to use agents safely and effectively.
• Protect against new threats, including those emerging from the agentic era.

Feedzai also raised open questions that come with greater autonomy: what an agent should do, where the human should stay in the loop, and what guardrails are needed. The source does not give pricing or availability details for Farol.

Trust as the thread through the day

The event brought together customers, partners and professionals from banking, fintech, payments, technology and the wider fraud ecosystem. Feedzai CEO Nuno Sebastiao opened with three words: technology, adoption and trust. The company's argument is that capability alone does not create transformation. Adoption has to follow, and people have to trust what the technology is doing.

Speed was one early theme. Astronaut Tim Peake described how astronauts prepare for high-stakes decisions through virtual reality rehearsals, stress tests and detailed procedures. "Failure is the best teacher," he said, according to Feedzai. The company drew a parallel to financial crime, where real-time detection is a constant goal but, in its words, speed without good information is just a faster route to the wrong decision.

Sean Hegarty of JPMorganChase described building a digital bank in 420 working days. Fraud controls were layered from the start across onboarding, login and transactions, so that if one control is beaten, another still provides protection. His lessons included getting the design right up front, thinking about end-to-end ownership of fraud, and communicating early.

Proportionate friction and signal sharing

Customer discussions questioned whether zero friction should be the goal when money or identity is at risk. One strong theme was proportionate friction. If someone is about to pay while under the influence of a scammer, a pause or a targeted warning can help. A single question informed by the payment, the payee and the customer's circumstances could be enough to break the spell. Participants used the phrase "unsticking" the customer from the scam.

Feedzai also reported a call to build risk around the customer rather than the function. Identity, authentication, behavior, payments and fraud often sit in different parts of an organization, while customers only see their bank. A more customer-centric approach means better data, better customer profiles and more dynamic decisions instead of relying solely on static rules and thresholds.

The third theme was that fraud is a network problem. One line from the law enforcement perspective was "Criminals have a network. We don't." A scam can start on an online platform, pass through a telco, reach a customer and move through a mule account, so no single institution sees the whole picture. Speakers from Stop Scams UK, Cifas, UK Finance and City of London Police agreed that no single team, institution or sector can solve it alone.

Feedzai highlighted a shift in language from reactive data sharing to signal sharing, since more data does not guarantee the right signal reaches the right place at the right moment. One discussion used the word "target" rather than "victim," because anyone can be targeted.

What it means for fraud teams

Feedzai says risk teams face rising workloads and fragmented data while fraud and financial crime grow more complex. Farol is its answer to that pressure, with agents taking on research, investigation and ruleset work. The company stresses that human oversight and guardrails remain central.

Its closing message was that better decisions, proportionate friction and connected networks matter most, with trust underneath all three.