Fraud control that shows its work.
Loci is real-time transaction monitoring and fraud detection for banks and fintechs. Stop losses before money moves, cut the false positives burying your analysts, and hand auditors an evidence trail instead of a black box.
Fewer losses. Fewer false alerts. Faster answers.
Fraud platforms get judged on four numbers: what you stop, what you wrongly flag, how fast you adapt, and whether you can defend it all later. Loci is built to move each one.
Weighted, context-aware signals replace binary thresholds, so one noisy check stops flooding your alert queue and your customers stop getting blocked for behaving normally.
Continuous device, location, and behavioral risk scoring catches ATO the moment a session stops looking like your customer, before the transfer, not after the complaint.
When a new fraud typology hits, describe it in plain language, shadow-test it against live traffic, and deploy, with no engineering ticket, no release cycle.
Every score decomposes into named signals with evidence attached. The lineage your auditors and regulators ask for is a by-product, not a project.
One control layer for the whole customer journey.
From the first signup to every payment and screening check, the same signals, the same evidence, the same explainable decisions.
Approve more of the right customers.
AccessGate can capture device, session, and behavioral signals from early onboarding moments. Catch synthetic identities and device reuse at signup without adding friction for real customers.
- Device + behavioral checks during signup
- Device-reuse detection across accounts
- Risk score returned before account creation
{
"session_id": "sess_8f3a…",
"device_fingerprint": "fp_c41d…",
"device_seen_on_accounts": 4,
"typing_cadence": "scripted",
"recommendation": "step_up_verification"
}
Stop account takeover before money moves.
Authentication shouldn't end at login. AccessGate scores the session continuously, device, location velocity, and behavioral drift, so account takeover surfaces the moment behavior stops looking like your customer.
- Impossible-travel and location velocity analysis
- Behavioral biometrics: mouse, keystroke, navigation
- Session risk feeds transaction decisions directly
{
"risk_score": 98,
"velocity_kmh": 102425.40,
"travel_feasibility": "impossible",
"risk_factors": [
"Travel velocity exceeds the speed
of a commercial aircraft."
]
}
Score every transaction in real time.
Loci's decision engine is benchmarked to evaluate active models against each transaction in low-latency paths and returns a decision with per-signal contributions, not a black-box score. Analysts see exactly why a payment was flagged before they touch it.
- Decision, score, and triggered logic in one response
- Velocity, amount, beneficiary, and channel controls
- Shadow-test new controls on live traffic safely
{
"decision": "review",
"fraud_score": 47,
"explanations": [
{ "rule": "LIFECYCLE_COMPOSITE_001",
"contribution": 9.65 },
{ "rule": "High Amount Withdrawal Check",
"contribution": 6.95 }
]
}
Meet AML obligations without drowning in alerts.
With AML screening enabled, screen against global watchlists such as OFAC, UN, EU, and UK lists, with PEP checks and relationship detection that flags customers connected to sanctioned entities, not just direct matches. Every hit carries citations.
- 6+ daily-updated global watchlists
- Relationship detection for indirect exposure
- Investigation-grade entity lookup with citations
{
"match_type": "relationship",
"lists": ["OFAC", "EU"],
"connection": "director of sanctioned entity",
"citations": 3,
"queue": "relationship_review"
}
The system behind the outcomes.
Each part has a name and a job, discover patterns, decide in real time, investigate with evidence, improve. You buy the outcomes; this is how they're produced.
Loci's decision engine evaluates every active model against each event in real time and returns the decision, fraud score, and per-signal contributions, with 21–49ms p95 latency in benchmark testing.
Describe what risky looks like in plain language. FLM turns it into weighted, versioned detection models, signals that accumulate, not brittle if-then logic, shadow-tested before going live.
Mines your historical data for fraud patterns and drafts evidence-scored candidate controls, each with estimated capture and alert load shown upfront.
Device fingerprints, behavioral biometrics, and location risk at every checkpoint, login, onboarding, payments, and account recovery.
Sanctions and PEP screening against 6+ daily-updated global watchlists, with relationship detection and citation-backed entity lookup.
Alert queues and case workflows that link evidence across transactions, and when enabled, sessions, devices, and screening results, so investigations start with context.
Cut false positives without sacrificing capture.
Autographer mines your own historical data for the patterns your current models miss, and hands your team review-ready controls with estimated capture and alert load shown before shadow testing.
amount > 50000 ∧ channel = AGENT
beneficiary_is_new ∧ count_1h > 3
| Control candidate | Estimated capture | Estimated alert rate | Evidence |
|---|---|---|---|
| Block cards with prior chargeback history | 87% | 0% | see evidence → |
| Flag high-risk email domains | 63% | 2.1% | see evidence → |
| Review orders > $10k from high-risk regions | 41% | 4.3% | see evidence → |
The platform that sees the full picture.
Every layer feeds the next. Signals become decisions, decisions become cases, and analyst feedback flows back into better controls, with governed human approval for production changes.
One POST, one explained decision.
Submit transaction and event payloads, connect session and behavioral risk signals, and receive real-time decisions with the evidence needed for action and review.
Read API Docs{
"transaction_id": "TX900000600144",
"entity_id": "E00018",
"amount": 8450.00,
"currency": "USD",
"source_account_number": "1231231231",
"beneficiary_account_number": "7100000006",
...
}{
"success": true,
"decision": "review",
"fraud_score": 20,
"explanations": [
{
"rule": "LIFECYCLE_COMPOSITE_001",
"triggered": true,
"contribution": 9.65,
"reason": "Lifecycle anomaly detected."
},
{
"rule": "High Amount Withdrawal Check",
"triggered": true,
"contribution": 6.95,
"reason": "Amount is suspiciously high."
}
],
...
}Discover. Decide. Improve.
From customer activity to governed decisioning in one operating loop.
Ingest & Contextualize
Send transaction, session, device, and behavioral events through APIs, with custom fields, reference tables, and historical context.
Evaluate & Explain
MADIE evaluates active controls and returns a decision, score, triggered logic, and evidence for review or action.
Test & Improve
Shadow-test new controls, feed analyst outcomes back in, and let Autographer surface the patterns your models are missing.
Unify fraud signals before losses move.
Bring transactions, sessions, devices, and behavior into one explainable control layer, and see it running on your own data.