Feedzai ships Farol agent inside RiskOps Studio to cut fraud investigation time

Feedzai launched Farol, an AI agent embedded in RiskOps Studio that helps fraud teams analyze rule performance, summarize investigations, answer product questions, and draft Suspicious Activity Reports (SARs) — inside each institution's own environment with audit trails.

FINANCE desk — banks keep piloting chatbots that cannot see live case data; an agent that already sits inside the fraud stack can cut investigation busywork without exporting customer data to a separate tool.

What the page says the launch is. The subhead says an embedded AI agent brings autonomous reasoning directly into core workflows, eliminating the integration gap stalling financial institution AI pilots. The lede says Feedzai, the global leader in AI-native financial crime prevention, today launched Farol, an embedded AI agent engineered to slash fraud investigation latency and sharpen threat detection across risk operations in banking. Global leader is the company’s claim. This desk did not rank vendors. Latency, here, is how long an investigation takes. These lines are the release’s.

The pressure the release names, and where Farol sits. The launch lands as fraud grows costlier to fight on every front. Mandatory reimbursement rules are increasing loss exposure, while technology spend and manual review costs keep climbing. Farol is built to alleviate those pressures. The release calls it an always-on agent that lets fraud teams do more without adding resources, fundamentally changing how they work. Integrated directly into Feedzai’s RiskOps Studio, Farol delivers conversational intelligence, automated rule performance analytics, and contextual case summarizations, all within the software stack analysts already use. RiskOps, in plain words, is the risk-operations workspace where fraud and anti-money-laundering analysts work cases. That gloss is this desk’s. The release prints the product name. Always-on, and do more without adding resources, are the company’s aims. This desk did not staff a team or time a case.

The 68% line, and whose it is. Feedzai’s agentic AI work is aimed at a bottleneck in institutional risk management. While 68% of financial institutions are actively testing agentic AI, according to Feedzai research, most deployments fail to yield operational efficiencies because third-party AI models operate in isolation from real-time transaction data. Farol is supposed to solve that split by embedding directly into existing risk workflows, in order to deliver full context, transparent audit trails, and autonomous execution capabilities without disrupting core infrastructure. 68% is a Feedzai research claim. The page does not name the survey, the sample, or the year of that research. Agentic, here, means the software can take a series of steps, not only answer one question. That gloss is this desk’s. An audit trail is a record of what the agent looked at and what it suggested. Full context, transparent audit trails, and autonomous execution are the release’s claims. This desk did not read the research note, and it did not watch an agent act on a live payment.

The SEB quote, as a quote. Justinas Rekus, Fraud Prevention Business Owner at SEB, said having a single, intelligent interface to handle data retrieval, insight generation, and production-ready rule suggestions fundamentally transforms how they refine their fraud strategies. The release calls SEB a leading northern European financial services group with an international presence in more than 20 countries worldwide. Leading, and more than 20 countries, are the release’s description. This desk did not interview him, and it did not count SEB’s countries.

The four skills the release lists, and only those. Farol launches with purpose-built skills. Risk Strategy interrogates and analyzes rule sets to surface actionable insights, including identifying rules that are generating noise without catching fraud, and recommending sharper thresholds to enhance performance. Management of rule hygiene becomes a task that takes minutes and not days. A noisy rule, in plain words, is one that fires often and does not find fraud. Rule hygiene is the work of cleaning that set. Investigations retrieves and summarizes alert data in moments, giving analysts the context they need to work cases faster and with greater conviction, proven to reduce alert handling times by 20%. Twenty percent is about one fifth. Proven is the release’s word. It is not a stopwatch this desk held. Knowledge is an always-on product expert built into the workflow. Users can ask Farol how to do anything on the platform and get the answer instantly, without leaving the user interface. SAR Drafting drafts Suspicious Activity Reports up to 12x faster by cutting the manual effort of compiling and summarizing information. A SAR, a Suspicious Activity Report, is the report a bank files with regulators when it suspects financial crime. Up to 12x means as much as twelve times as fast. Up to is the release’s ceiling. Minutes and not days is the Risk Strategy clock. It is not the 12x line, and it is not a 13x line. The wire does not print 13x.

The IDC quote, as a quote. Sam Abadir, research director for risk, financial crime, and compliance at IDC, said fraud investigation is one of the clearest near-term use cases for agentic AI. He said an agent with direct access to case data and analyst tools removes the manual work of assembling that data, driving faster resolution without adding headcount. Longer term, he said, agents that flag which rules generate noise without catching fraud let teams tune detection logic continuously, not just during periodic reviews. Near-term and longer-term are his split. Headcount is his word for not hiring more people to do the assembly. This desk did not interview him, and it did not audit IDC.

Where the insights stay. The release says Farol builds on Feedzai’s track record, from operationalizing machine learning and explainable model outcomes to real-time cross-channel risk decisioning and federated network intelligence. Those are the company’s history lines. This desk did not audit that history. Feedzai Farol operates within each financial institution’s own environment, so the information and insights it surfaces stay inside the customer’s estate, multiplying the impact of fraud teams and helping them solve the problems that matter most. Own environment, and the customer’s estate, mean the bank’s own systems. The insights are not described as an export to a separate chat tool. Multiplying the impact is the release’s claim. This desk did not measure a team’s output.

The product chief, as a quote. Pedro Barata, chief product officer at Feedzai, said fraud teams don’t need another tool to check. They need an agent that already knows their environment and the problems they want to solve. He said the Farol agent puts actionable intelligence directly in the hands of the people fighting fraud every day. It surfaces the right information faster, takes repetitive work off their plate, and gives analysts the speed and confidence to act, so banks can focus on keeping their customers safe. Another tool to check is his contrast. It is the same idea as the own-environment sentence. It is not a customer count. This desk did not interview him.

What the about box says the company is. Feedzai says it powers trust in global finance. It protects people and payments using what it calls trusted AI to detect and prevent financial crime, fraud, and money laundering in real time, so money moves safely. Every year, the world’s top banks, payment networks, and acquirers use Feedzai’s technology to safeguard more than one billion consumers and $9 trillion in payment volume. World’s top is the company’s claim. More than one billion is the consumer floor. $9 trillion is the payment-volume figure. The page does not print a year for that volume beyond “every year.” The site it names is feedzai.com. The learn-more line points at www.feedzai.com/farol-agentic-riskops. SOURCE Feedzai is the last line. These lines are the about box. This desk did not audit the payment volume.

What the product page adds, and what it does not let you file as the wire. The Farol page at feedzai.com/farol-agentic-riskops, read the same day, is not the 09:00 ET release. Its Risk Strategy block says Farol helps a team optimize risk strategies 13x faster than they do today, and that analysis that used to take hours now takes minutes. 13x, and hours to minutes, are that page’s. They are not sentences on the CNW wire. The wire’s Risk Strategy clock is minutes and not days. Do not collapse them. The page’s Investigations block says review alerts up to 20% faster. A later answer on the same page says Farol accelerates investigations by up to 20%. The wire says proven to reduce alert handling times by 20%, without “up to.” Up to is the product page’s hedge. Proven is the wire’s word. The page says filing a SAR can be 12x faster, and a later line says file SARs 12x faster. The wire says up to 12x. A question on the page asks whether a bank can connect its own AI agents. The answer says Bring Your Own Agent is an upcoming Agentic RiskOps capability, still being developed, and that the company is looking for customers to test it. Upcoming is not shipped. Do not file a customer’s own agent as live. Another answer says Feedzai’s agentic AI is built to augment analysts, not replace them. Agents surface insights and recommendations, with transparent reasoning, so analysts can verify outcomes and take actions as they see fit. A third answer says Farol is embedded directly into Feedzai RiskOps and the workflows teams already use, rather than adding another disconnected tool. These lines are the product page’s. This desk did not request a test of Bring Your Own Agent.

What the picture shows, and what the words do not. The card is Feedzai’s Farol launch slide. A pale green field holds a RiskOps Studio window. The chrome says Production, Tenant 01, and the initials DR. The left side is a Digital Activity list marked 38, with columns for amount, risk score, status, assignee, and queue. Three rows are readable: $2,850.50, Medium, Pending Review, Ana Blanchard, To Review; $1,540.00, Medium, Pending Review, Unassigned, To Review; $890.75, Low, Resolved, Daniela Pereira, Transfers Alerts. The right side is a Farol chat. The question on the slide says the team needs to reduce alerts by up to 50 per day, and asks which rules have triggered at least 20 times but have not found any fraud in the last 30 days. Farol’s answer on the slide says three rules have triggered repeatedly but found no confirmed frauds in the last 30 days, and names R_0001_medium_risk_model_score, R_0014_overseas_login, and R_0018_international_high_value. Those names, the dollar amounts, the people, the 50-a-day line, the 20 times, and the 30 days are the graphic’s. The 24 Sep release does not print them. Do not treat the still as a case this desk opened, and do not treat 50 alerts as the wire’s 20% claim. The slide has no date text.

Plain English for the rest of the card. Farol is an AI agent inside Feedzai’s RiskOps Studio, the workspace fraud analysts already use. A SAR is a Suspicious Activity Report filed to regulators on suspected financial crime. A noisy rule fires often and does not catch fraud. 68% is Feedzai’s research claim about institutions testing agentic AI. 20% is the wire’s alert-handling claim. Up to 12x is the wire’s SAR-drafting ceiling. Minutes and not days is the wire’s rule-hygiene clock. 13x, and hours to minutes, are the product page’s Risk Strategy lines, not the wire’s. Bring Your Own Agent is upcoming on that page, not a skill the wire says shipped. One billion consumers and $9 trillion are the about box. 09:00 ET is 9:00 a.m. Eastern and 1:00 p.m. UTC. The schema’s published and modified timestamps are the same minute. This filing is the 24 Sep announcement.

PRIMARY here: Feedzai’s 24 Sep 2026 CNW Newswire release, stamped Sep 24, 2026, 09:00 ET and datelined New York and Lisbon — Tier A PRIMARY, the company’s own announcement. The Farol launch, the subhead, the global-leader claim, the latency and detection sentence, the reimbursement and cost sentences, the always-on and no-added-resources lines, the RiskOps Studio integration, the conversational intelligence and case-summarization list, the 68% Feedzai research claim, the isolation-from-transaction-data sentence, the full-context and audit-trail and autonomous-execution sentence, the Rekus quote and the SEB description, the four skills, the minutes-and-not-days line, the proven 20% alert-handling line, the up-to-12x SAR line, the Abadir quote, the track-record sentence, the own-environment and customer-estate sentence, the Barata quote, the one-billion and $9 trillion about-box lines, and feedzai.com are that release’s. The 13x line, the hours-to-minutes line, the product page’s “up to 20%,” the SAR lines without “up to,” and the upcoming Bring Your Own Agent answer are the product page’s. The rule names, the dollar amounts, and the 50-alerts line are the launch slide’s. NOT claimed: that this desk opened RiskOps Studio, timed an alert, or filed a SAR, that 68% was independently surveyed, a named sample or year for that research, that proven 20% or up to 12x was audited, that 13x is a sentence on the wire, that Bring Your Own Agent has shipped, that the people and amounts on the slide are a live case, a dollar price for Farol, a stock tip, or investment advice. Distinct from the already-filed dun-bradstreet-db-ai, place-maxwell-ai-mortgage, and microsoft-eviltokens.