Colibrix is moving from iGaming into subscription, and chargeback ratios are climbing on the new merchant book. Fraudio is the AI, scheme-monitoring and merchant-oversight platform built for exactly that transition — one centralised brain trained on 2M+ merchants, with the deepest bench in the industry on subscription-merchant patterns and both Visa VAMP and Mastercard SMMP.
Colibrix holds EMI licences in the UK and Malta, is a principal Visa and Mastercard member, and is repositioning its acquiring book from iGaming into subscription and recurring traffic. The strategic bet is sound. The gap Yulia's chargeback team and Sergejs' risk team both flagged is that the current stack has no specialised system for the pattern shift — chargeback ratios climbing on the new subscription merchants, no transaction-monitoring platform for the risk team, and Visa's VAMP still counting refunds against you. This is the layer Fraudio is built to be.
Fraudio spun out of an ING acquirer in 2019 and was built as an AI/ML company from first principles — not a rules engine that added ML later. The differentiator is the centralised database: acquiring data, issuing data and account-to-account data mapped into one homogeneous schema, so every model learns from the full picture instead of a single tenant's slice.
Feature enrichment at scale. Colibrix sends us the 16-80 data fields available on a transaction. Fraudio enriches each one with thousands of contextual features — interactions, aggregations over sliding windows, peer-group deltas, historical sequences. That's what turns "one transaction" into a decision the AI can actually make in sub-100ms.
Rules and AI in one pipeline. Rules handle compliance and explicit business appetite (scheme mandates, MCC blocks, jurisdictional limits). AI handles the pattern learning that no static rule can — with continuous retraining on feedback from every merchant in the consortium.
Privacy by design. Your data never leaks to another tenant. The shared benefit is model performance, not raw information — data segregation is logical by default, physical on request.
Fraudio's centralised dataset spans a genuinely global merchant and issuing base. When Colibrix's subscription merchants transact across borders, the model has already seen behavioural analogues in the equivalent MCCs, currencies and channels somewhere in the consortium. That's the network effect that separates a real AI platform from an ML product marketed as one.
Visa's Acquirer Monitoring Program and Mastercard's Standards for Merchant Monitoring Program are how the schemes hold Colibrix accountable for what happens on your BINs. Fraudio doesn't approximate them — the platform natively tracks VAMP and SMMP at merchant, sub-portfolio and BIN level, so your chargeback and risk teams see what Visa and Mastercard will use against you, before they use it.
Continuous tracking of Visa's combined fraud and dispute ratios. Early warning when a merchant or sub-portfolio drifts toward Standard or Excessive thresholds — with the drivers surfaced.
Mastercard's Standards for Merchant Monitoring tracked the same way — chargeback ratios, fraud ratios, excessive-chargeback merchants and high-risk MCCs flagged before they become a scheme inquiry or BIN-level order.
Why this matters for the subscription pivot: Visa's VAMP doesn't forgive you for refunding. Every refunded fraudulent transaction still counts. In subscription, where the friendly-fraud and dunning-refund rate is structurally higher than iGaming, VAMP is the metric that turns a strong quarter into a scheme conversation. Fraudio catches the merchants driving it before Visa does.
Subscription looks less risky than iGaming on the surface. The fraud shape is different — and it's the shape most acquirers under-instrument for on the way in. Fraudio's dataset includes deep subscription coverage across the consortium, and the models are pre-trained on the specific patterns your new book will hit.
Customers charging back subscription renewals they made months ago. Flagged by behavioural signatures long before the chargeback arrives.
Failed-payment retry loops that generate scheme complaints and load VAMP without ever settling — modelled by velocity, decline-reason clustering and retry-cascade detection.
Attackers probing tokens on subscription flows. Detected as coordinated bursts across cardholders and channels — the kind of pattern only a centralised dataset actually sees.
Merchant collusion patterns specific to subscription — shared emails, transaction spikes, abnormal cross-merchant links. Visualised natively in the platform.
Refunds still count against VAMP. The system flags merchants generating refund velocity that will drift them toward scheme thresholds before it happens.
Web-extracted merchant intelligence flags subscription merchants selling outside their declared MCC — the compliance issue no chargeback tool alone will surface.
The platform is what turns the AI into a workflow. Sub-tenancy for sub-processors and merchants, role-based access for AML analysts, compliance managers and risk investigators, and every capability the team went through in the deep dive.
Customisable KPIs on the executive dashboard. Drill into merchants, transactions, alerts, cases — for the head of chargebacks and the head of risk at the same time.
Give portal access to sub-processors, merchants and internal teams under separate tenants. AML analysts, compliance managers and risk teams each see the surface they should.
Screening rules (real-time authorisation) and monitoring rules (entity-level pipelines). The audit function warns you if a rule is logically flawed. Shadow mode tests every rule on historical data before it goes live.
Generate rules from natural-language descriptions. "Block subscription merchants with dispute ratio above X in the last 7 days" turns into a live rule your team can review and enable.
Centralised queue of rule- and AI-triggered alerts. Analysts mark false positives, add comments, escalate, run full investigations and export outcomes. Every action audit-logged.
Graph views of merchants sharing emails, devices, IPs, or exhibiting abnormal transaction spikes — the collusive-merchant pattern the risk team went in specifically to see.
Merchants enriched with web-extracted information — smart summaries flagging potential forbidden-goods sales, MCC mismatches and scheme-standard failures that a payment-data-only view will miss.
Proxy and VPN detection, device linking behind IPs, cross-linking of phone, email and IP where the data fields are provided. Integrated with the gateways and processors Colibrix runs.
Full compliance stack — KYC, KYB, PEP, sanctions, adverse media — with support for complex UBO structures and offshore jurisdictions, local providers plugged in where needed.
Ingest chargeback files from Silverflow, your current processor and any future partner — into one ratio, one merchant view, one investigation queue.
Full API reference, integration guide, data schema, event handling and webhook specification — hosted at docs.fraudio.com. The team can start scoping the field mapping and the technical assessment immediately.
API reference, integration guide, data schema, webhook specs and sample payloads.
Fraudio prevents chargebacks upstream. For the ones that already landed, we work with a specialised chargeback-management partner that connects directly to the card schemes and fights winnable disputes on a pure-performance model — they only charge when they win. Their system integrates back into the Fraudio portal for real-time status.
The result for Yulia's team: Fraudio reduces the volume of chargebacks the team has to handle. The partner recovers the ones that shouldn't have landed. Colibrix pays only for wins.
Your subscription book gets the network effect of 2M+ merchants and issuing + acquiring data from transaction one — no cold-start period while your own data seasons.
Both scheme programmes are native to the platform, not bolted on. Colibrix sees the ratio drift as it develops — with the driver merchants surfaced, not buried.
ISO 27001 certified, GDPR compliant, ING-acquirer heritage. The compliance shape the FCA (UK) and MFSA (Malta) both already expect from a vendor.
Three-year contract with an out-clause in the first six months. Term-based discounts. Pricing tied to volume, product mix and team size — no hidden overage.
Viva Wallet is a European payment institution and acquirer that deployed Fraudio to score card fraud in real time and monitor merchants for fraud and money laundering. It scaled the merchant book past half a million without scaling the fraud team. The operational shape Colibrix's subscription pivot needs.
“Fraudio enables us to detect fraudulent merchants and money laundering, ensuring the safety of our clients against fraud in payments. This has been extremely helpful to our growth by providing us the ability to focus our efforts in a much more accurate manner.”
A single-page summary of what makes Fraudio different — the centralised AI, the network effect and the customer results — for the wider Colibrix team.
