Prepared for
PayProcc · Jean-Paul Schmitz
Everything from the intro: how the platform plugs into Akurateco without an integration project, how it handles the specific fraud patterns Jean-Paul described (burst attacks between 1-2am, VPS/IP rotation, rotating email, card batch testing, simulated-trading rings), the three-month pilot shape, the 15-minute demo, and the pack for the deep-dive with Robin.
The three API connections to the Akurateco stack — pre-authorisation, post-authorisation, and post-event (chargebacks and fraud notifications) — already exist in production. For PayProcc, turning Fraudio on is a configuration on the Akurateco side plus a Fraudio account provisioning. Not a 3-6 month integration project. Live within days, pay for what you score.
The pilot shape we agreed, laid out month by month. The commercial frame underneath: 2-year agreement with a 3-month pilot built in, so PayProcc keeps optionality through the measurement window.
Fraudio account provisioned on the Akurateco connector. Historical data ingested. Shadow-mode scoring from day one on the targeted MCCs and merchants. Baseline measured against current VAMP ratios.
Tune the operating point (red / yellow / green thresholds), add explicit rules for the specific patterns (burst detection, IP/VPS, email rotation). Measure fraud ratio reduction and approval-rate impact weekly.
Clean yes/no decision at month 3 based on the measured impact on the spiking segments. The pilot sits inside the 2-year agreement, so PayProcc keeps optionality through the measurement window.
Behavioural burst detection on the resolved entity (IP, merchant, BIN) with time-of-day context as a feature. Transaction scores escalate yellow → red across sequential attempts in the window. Explicit rules on top for the specific IP ranges and time windows.
IP, device and VPS-range features are resolved to the attacking entity in real time, not treated as independent transactions. Rotation itself becomes the signal. Proxy and anonymous-IP detection confirmed as a platform capability.
Entity resolution ties the rotating email, device and session back to a single actor. Batch-testing patterns (same BIN, same IP, high failure rate in a tight window) are the exact use case the AI and rules are tuned for.
Centralised consortium view across every merchant on the network means fraud rings active in one trading provider inform scoring on every other. The ring is seen before it finishes attacking PayProcc.
You mentioned you're deciding between MaxMind and Fraudio by end of week. The honest architectural distinction, not a sales take.
Full side-by-side write-up: Fraudio vs MaxMind →
You asked for actionable output rather than raw 0-1 risk scores. Fraudio returns three decisions, with reason codes, that drop straight into an operations queue.
Decline the transaction. High-confidence fraud, with reason codes. Measured impact on fraud ratio captured in real time.
Queue for ops review. Ambiguous but worth a human check. Case workspace with SLA tracking per analyst.
Approve. Clean transaction. Measured against approval-rate baseline so pilot can verify no harm done to legitimate volume.
You flagged that non-EU transactions from India, Malaysia, Singapore and parts of South America often lack enforced 3DS, which leaves a visible gap in the current stack. These are precisely the geographies where Fraudio's centralised AI carries the most value — because the fraud patterns from these corridors are already visible across the consortium.
The AI treats origin geography as a feature, not an exclusion. Non-3DS transactions from these corridors are scored with the same model that has seen their behavioural signature on every other acquirer and PSP running on Fraudio.
Guided walkthrough of the full platform — real-time scoring, case management, rule editor, recommendations, dashboards. This is the same surface you'll see live in the deep-dive with Robin.
Everything referenced on the call, in one place, for PayProcc's internal review ahead of the technical deep-dive.
Guided walkthrough of real-time scoring, case management, rule editor and dashboards.
Side-by-side architectural comparison — for the end-of-week decision.
Executive overview: centralised AI, network effect, platform scale, customer results.
Supervised + unsupervised + rules, feature engineering, reason codes, model governance.
AML transaction monitoring, typology coverage, case management, SAR / STR export.
Onboarding screening architecture, data providers, UBO resolution, rescreening.
Visa VAMP and Mastercard SMMP coverage: tracking, dashboards, breach prevention.
Platform security, data handling, residency, cryptographic controls, privacy architecture.
Multi-region active-active, failover posture, RPO/RTO, backup regime.
Current ISO 27001 certification, scope, auditor, validity.
Full API reference, integration spec, data schema, webhooks — for Robin's deep-dive.
Grab a slot for the technical deep-dive with Fraudio's co-founder Robin.
Product and data technicalities with the person who built them. Bring the engineering questions, the specific fraud patterns, and anything the demo didn't cover.
Book the deep-dive