SENTR · FOR FINTECHS & NEOBANKS
Keep genuine customers moving while your fraud controls adapt. SENTR connects account activity and money movement with configurable rules, machine learning and AI explanations—giving your team the context to investigate and the controls to respond.
EVALUATION QUESTIONS · ILLUSTRATIVE, NOT CUSTOMER TESTIMONIALS
HEAD OF FRAUD / RISK DIRECTOR
"Our false positive rate looks fine on paper. It doesn't look fine when you see the activation dropout it's causing."
A signup decision should not use the same assumptions as a withdrawal. Event-scoped rules and policies let your operator express that distinction.
CFO / FINANCE DIRECTOR
"We're spending more on manual review than on the fraud platform itself. And the platform is still missing things."
Measure losses, review effort and blocked good customers separately. The pricing and value model helps make those assumptions inspectable.
CTO / HEAD OF ENGINEERING
"We have the data to build a better model. We don't have the loop to feed it back."
Customer-specific anomaly detection trains overnight on clean history. Validated investigation feedback supports learning, alongside governed changes to rules and policies.
SENTR gives growing fintechs a connected fraud workflow: ingest the events, score the risk, explain the decision, investigate and improve. Explore the controls your team can put to work.
From onboarding friction to payout risk, SENTR gives your team one place to configure decisions and understand their consequences.
Store rule and model attribution, configuration changes and reasoned overrides on a connected record. Give operations, risk and engineering the evidence to understand a decision without reconstructing it from separate systems.
A blocked legitimate payment can undo the experience your product team built. Inspect noisy rules, test candidate changes and use customer context to tune risk appetite. Evaluate both fraud outcomes and customer friction before rolling changes into production.
Use policy to automate clear decisions and route uncertain activity into review with its explanation and history. Queue and rule analytics help you see where the team spends time—and which changes deserve testing.
New accounts, changed devices and unusual payout patterns mean more when viewed together. Map your event types and identifiers into SENTR, then configure rules for your use case. Credit underwriting and AML screening are not being offered here.
The shared SENTR loop applied to your team’s work: Ingest, Detect, Decide, Explain, Investigate and Improve. See how the loop works →
Slide to explore all six stages
Stage 1
Map BNPL, wallet and payment events with the account and device history your app already holds.
Open →Stage 2
Rules and scoring on mapped events — including velocity and linked-entity patterns you can investigate.
Open →Stage 3
Scoped policies and scoring profiles with deterministic explanations your legal and risk teams can review.
Open →Stage 4
Readable attribution and decision evidence for compliance review and partner questions.
Open →Stage 5
Queues and cases with device, session and payment context — overrides require a written reason.
Open →Stage 6
Confirmed outcomes feed lists, policy review and operator-led improvements on the shared engine.
Open →Every confirmed outcome is a chance to improve. Customer-specific anomaly detection learns from clean history; operators use investigation feedback to review rules, lists and policy. That is the loop—not a promise of automatic dispute wins.
GOVERNANCE & EVIDENCE
Risk, engineering and compliance should work from the same decision record. SENTR stores attribution, configuration changes and human overrides for review. Applicable legal obligations depend on your business and deployment; the platform provides evidence, not certification.
Read the EU AI Act notes →Decision records support your assessment. Regulatory classification and compliance remain matters for your organisation and qualified counsel.
Put SENTR beside your current controls for a qualified, free 50-day read-only evaluation. Compare decisions, investigate differences and separate confirmed findings from modelled impact. Your data. Your decision.
Connect · agreed scope
We map the agreed event types, identifiers and available history with your technical contact. Your existing system keeps making production decisions. Integration scope and readiness determine the start date.
Evaluate (Days 1–50)
SENTR scores the agreed feed without enforcing its decisions in production. Compare outcomes, investigate disagreements and record confirmed findings separately from open questions.
Decide (Day 50)
Review a Day 50 report covering decision comparisons, investigated findings, operational measurements and modelled financial impact. Some fraud and chargeback outcomes need longer to mature. Decide together whether to proceed, refine the evaluation or stop.
Shadow mode is qualified and read-only while evaluation runs. No production commitment required to evaluate. You choose whether Shadow Mode is the right evaluation path.
How Shadow Mode works → Book an Architecture Session →
FOR YOUR ENGINEERING TEAM
Use supported REST APIs, webhooks and other mapped ingestion paths. Scoped credentials and a provisioned sandbox support integration testing. Automated Adyen/Stripe mapping is still in development.
Agree the event map, history requirements, test cases and production boundary before implementation. Explore data & integrations →
SENTR uses AWS regions in Ireland and Germany. Confirm data flows, explanation-processing subprocessors and residency terms in the security review and DPA. Read security & data →
Keep your payment infrastructure while evaluating SENTR. Agree data export, retention and exit arrangements before production; we do not assume a change of processor is required.
Inside the workflow · illustrative configurations
An identity check, a trusted account and a familiar device each answer different questions. Bring them together without treating one successful check as a permanent exemption.
An account and session connect to a new device and first-seen IP. Inspect that change alongside authentication history and the next withdrawal or transfer. Neither a new device nor a new IP proves takeover; event-specific controls and investigation determine the response.
Ingest registration, supplied KYC results, stable account IDs and available device/IP history. Entity import and backfill help distinguish an empty record from genuinely new activity.
Apply registration-scoped rules for the signals you have. Keep onboarding and money-movement thresholds separate. SENTR uses identity-provider results; it does not perform document verification itself.
Follow linked profiles and earlier cases, apply the configured checklist and record an outcome. Shared infrastructure may be legitimate; the link is a reason to investigate, not a verdict.
Connect authentication, account changes and withdrawal events with available session, destination and history fields. A payment allowlist should not erase new-session context.
Use distinct event policies and list scope. Return approve, review or block to the application; any access restriction, step-up or withdrawal hold is implemented by that integration.
Start from the account or session profile, inspect the sequence and record a reasoned decision. Keep takeover, legitimate access and uncertain findings distinct in the outcome taxonomy.
See the controls behind these workflows: SENTR.Citadel product deep dive → Compare guided SENTR.Tower →
Your industry shapes the rules. Both models share customer-specific machine learning and AI explanations. Choose how much configuration your team wants to own.
THE SHIFT
Every new product, payment method and market changes the risk. SENTR connects your event data, decision controls and investigation work so your team can adapt the system—not spend its day working around it. Bring us the part of your current operation that needs to change. We will show you how it would work, then agree how to test it.
Book your Architecture Session →A working session, not a pitch deck. Optional Shadow Mode is free for qualified SENTR.Citadel evaluations; production is a separate decision.