in fraud losses reported by consumers to the U.S. Federal Trade Commission during 2025
Investment scams were the largest reported loss category, at US$7.9 billion.Source: U.S. Federal Trade Commission · March 25, 2026Fraud prevention at the moment of transaction
Someone is about to send their savings. Your team can still help them pause.
YAP Sentinel is designed to help customer-facing transaction teams surface signs of coercion, connect the context, and route the case to an accountable human before an irreversible payment or transfer.
Trading Tool provides the transaction-control foundation today. Sentinel is the supervised analysis layer in controlled pilot. Consequential actions still require human approval.
The scale of the problem
The losses are counted in billions. The decision happens one person at a time.
Complaint systems show the scale of reported harm. Customer-facing teams encounter the moment when an intervention may still be possible: before the funds are sent.
in fraud losses reported to the Canadian Anti-Fraud Centre during 2025
The Canadian Anti-Fraud Centre received more than 112,000 fraud reports.Source: Canadian Anti-Fraud Centre · April 1, 2026of 2024 EEA credit-transfer fraud losses were borne by payment users
The ECB and EBA say this was mainly driven by scams that manipulated people into initiating the transfer themselves.Source: European Central Bank and European Banking Authority · December 15, 2025in reported losses associated with 2025 complaints involving cryptocurrency kiosks
IC3 received more than 13,400 complaints; the losses cannot be attributed solely to kiosks because some complaints included other transaction types.Source: FBI Internet Crime Complaint Center · May 15, 2026These figures come from separate complaint and industry-reporting systems. They should not be added together. They reflect reported or recorded harm—not the full scale of scams.
The gap after identity verification
The customer can be real. The intent can still be manipulated.
A person under a scammer’s control may present valid identification, know every answer, and insist that the transaction is voluntary. The risk often lives in the story around the payment: coached answers, secrecy, urgency, false profits, or one more fee required to release funds.
Sentinel is being designed to bring those signals into the same supervised case as the customer, transaction, wallet, and lawful public-source context—without treating any label or behavioral signal as proof.
One case, from first question to safer next step
Synthetic caseYS-0184
Before the money moves, there is still time to ask the right questions.
Follow one entirely synthetic case through the proposed Sentinel workflow. The software gathers and prepares; an authorized person decides what happens next.
A customer is told that one final fee will release investment profits shown on a trading platform.
The request sounds ordinary.
The customer wants to send funds to unlock a withdrawal. The recipient, wallet, and payment instructions came from someone met online.
Prepared case artifactStated purpose captured
The story reveals pressure.
Fake profits, a withdrawal fee, urgency, secrecy, and coached answers appear together. No single signal proves fraud; the combination warrants review.
Prepared case artifactBehavioral indicators separated from facts
Context becomes a case.
Indicators tied to the customer, transaction, wallet, and technical infrastructure are combined with findings from lawfully obtained public sources. Each item carries a source reference, timestamp, confidence level, and any unresolved questions.
Prepared case artifactEvidence synthesis prepared
Software stops. A person decides.
The case reaches a mandatory approval point. An authorized employee reviews the evidence and decides whether to continue, pause, or refuse the transaction; escalate the case; or prepare an external report.
Prepared case artifactHuman decision recorded
The customer gets a safer next step.
Staff receive calm, blame-free language that encourages a pause, evidence preservation, and support from trusted sources—without an accusation or a promise of recovery. Any improvement note is sanitized before human review.
Prepared case artifactSupport guidance prepared
This scenario is entirely synthetic and contains no customer or active-case data.
The product, as one supervised system
Operational controls underneath. Specialist analysis around the case. Human authority at the centre.
Sentinel does not replace the controls that already govern a transaction. It adds a coordinated analysis and evidence layer around them, then stops at the decisions that carry legal, financial, or human consequences.
Structure the transaction
Trading Tool supports adaptive information collection, identity workflows, screening signals, transaction gates, compliance queues, and evidence exports—with explicit limitations published alongside each capability.
Explore the transaction-control foundationMake the pressure visible
Eight specialist agents are designed to examine behavioral, public-source, on-chain, infrastructure, compliance, and customer-support context as one case—not eight disconnected answers.
Meet the eight supervised agentsStop for a person
Refusals, holds, filings, external reports, disclosures, third-party contact, and customer-facing accusations remain decisions for identifiable, authorized people.
See the approval boundariesProtection without shame
Our aim is simple: help more people keep their money.
People caught in scams are often acting under trust, fear, hope, or intense pressure. Sentinel is designed to help staff slow the moment down, explain the concern without blame, preserve useful evidence, and suggest practical safety steps. It never diagnoses, promises recovery, or recommends confronting a suspected scammer.
“You did the right thing by telling me. The safest next step is to pause before sending any more money.”
Example blame-free language prepared for staff review
Built for high-stakes transfers. Designed for review.