T-NEX Fraud Detection
T-NEX Fraud Detection is a transaction monitoring and case investigation module for fraud and anti money laundering teams at regulated institutions. It is part of the banking platform built by T-NEX GmbH of Horn-Bad Meinberg, Germany, which also lists an office in Kuala Lumpur. Active rules screen incoming transactions and group the resulting alerts into cases for each customer, with a dashboard, a case queue, investigation tools, rule backtesting and calibration.
Nine rule types cover three areas. Laundering indicators catch pass through transfers, structuring, threshold breaches, high risk countries and reactivated dormant accounts, a second set watches card and payment patterns, and a text based rule sends a transaction to the model. Thresholds and time windows are named parameters, and clear amount or time conditions run without a model call.
AI triage adds a risk score, a rationale and a recommended action, stamped with the model and prompt version, to help analysts prioritize. Analysts make the call, record a reason when they differ, and can draft a suspicious activity report for filing through the institution's own reporting system. If the AI service fails, the alert stays workable with a labeled rule based explanation and the analysis can run again later. The software is sold by project quote after a demonstration.
Capability Axes
Capability grades
15 of 15 axes rated · 7 graded A or B
AI triage sits on top of a rules engine that works on its own. Deterministic rules flag transactions on thresholds, time windows and customer context, alerts group into cases, and analysts investigate, backtest and calibrate without any model call. The model adds a risk score, rationale and suggested action for prioritization only, and when it is unavailable the alerts carry rule based explanations and work continues.
A qualified investigator decides every flagged case. Staff assign, escalate, close or reopen cases and record their reasoning when they depart from the recommendation, which carries its model and prompt version. When inference fails, a labeled rule based message keeps manual handling going, and the analysis can rerun once the service returns. Reports to authorities and payment decisions sit outside the investigation workflow.
Investigators see the originating rule, related transactions, the AI assessment and its reasoning, and the model and prompt version used. Rules can be backtested and calibrated on historical data, and failed model calls are flagged so the case proceeds on rule evidence. The score is labeled as a prioritization aid rather than proof of fraud or laundering. A benchmark or independent validation on real institutional data is not offered.
A 23 page Fraud Detection manual dated 30 August 2026, for application version 15.23.0, covers the nine rule types and walks through sample cases, and an institution can test the rules on its own historical data. Customer installations of the module, monitored transaction volumes and measured changes in detection or investigation time are not shown. The one installation on the evidence page, at an unnamed German institution, runs the Controlling product.
Transaction and customer context goes to a centrally configured AI service, with model choice, data flow, provider terms and access permissions set for each implementation, and training and demonstration environments run on sample records. Whether production inputs can train models, whether an institution can opt out, and how long inference data is kept are not stated.
Each transaction arrives with external identifiers, account references, amount, date, currency, channel, counterparty country and the customer's risk classification, and a technical assessment reviews what reaches the AI service and who can access it. Data flows, storage locations and participating providers are assessed for each environment, and privacy and hosting terms are agreed in the contract. The public privacy policy covers the website, and a retention schedule or subprocessor list for deployments is not public.
Roles, access permissions, an audit trail, backups, monitoring and change management are configured for each regulated installation, and the operating plan assigns updates and incident handling to named parties. A security certificate or audit report covering Fraud Detection and its AI service is not offered.
The module supports institutions' fraud and anti money laundering monitoring duties and keeps an alert, a review and a finding distinct. Analysts prepare a draft suspicious activity report, which the institution files through its own reporting system under its own legal process, with no automatic submission to authorities. T-NEX supplies the software rather than acting as a regulated institution.
Every recommendation is saved with its model and prompt version and the transactions behind it, and an investigator who reaches a different decision must state why. Institutions test rules on historical data and check fallback behavior before go live, and a separate assessment covers permissions, chosen models and data flow. Results of model recommendations across customer groups, or an independent study of them, are not shared.
An investigator can reopen a case, review the triggering transactions and record a manual decision when the AI suggestion is wrong, so the institution can correct its own handling. Reporting and enforcement decisions stay with the institution. An appeal route for an affected account holder, or vendor liability for a mistaken fraud recommendation, is not offered.
Fraud triage runs on the platform's central model service, and each result saves the model and prompt version used. The model, provider terms and data flow are chosen for each deployment, with GPU and compute capacity planned for the expected load. The model providers that process production case data are not named.
A defined data model covers transactions, accounts, cards and customer records, with identifiers for mapping and deduplication. The module shares the banking platform's data context and can reuse an institution's existing customer risk classifications, and interfaces and field mapping are set in the project. A ready connector to a named core banking system, or a direct filing link to a reporting authority, is not listed.
Institutions choose managed operation, where T-NEX arranges the infrastructure and support, or operation in their own cloud account, where they receive provider invoices and T-NEX is given access. Storage locations and participating providers are decided for each project, and a standard hosting region or an on premises installation is not offered.
Institutions start with a demonstration and a project assessment, and an editable procurement overview dated 11 September 2026 lists what can be configured, from data feeds and permissions to AI integration. Licensing follows the agreed installation and operating scope, and a subscription price, minimum charge or per transaction rate is not listed.
Fraud and anti money laundering analysts, and the specialists who own transaction monitoring, are the users at banks and other regulated institutions. Rules cover bank transfers, card and payment patterns and laundering indicators, and each case brings together the customer relationship and the records an analyst needs. Bank and payment workflows carry the most detail, with less on how the module fits insurers or other regulated firms.
Alternatives to T-NEX Fraud Detection
Other products that do the same kind of work as T-NEX Fraud Detection, and what each one does.
DataVisor unifies fraud detection, anti money laundering monitoring, customer and business onboarding checks, case management and risk decisioning on one platform covering the whole customer lifecycle.
Featurespace, a Visa solution since Visa completed its $946 million acquisition in December 2024, sells real time fraud and financial crime detection to banks, card processors, acquirers and credit unions.
Eastnets sells compliance and payments software to banks, payment firms, exchange houses and insurers.
NICE Actimize sells software that catches fraud and money laundering at banks and other financial institutions. Xceed puts fraud prevention, money laundering monitoring, OFAC screening and investigations on one AI platform for credit unions, community banks and smaller institutions.
Resistant AI sells document fraud detection and transaction monitoring to banks, lenders, payment companies and insurers.
Oscilar sells one platform for the risk decisions a bank or fintech makes about its customers: onboarding, fraud, anti money laundering compliance and credit underwriting.
Pricing
Vendor-published figures are labeled as such. Figures labeled “Estimated” are derived from third-party sources and have not been confirmed by the vendor.
| Entry Price | Pricing Basis | Data Protection Terms | Implementation | Source |
|---|---|---|---|---|
|
Project quote from T-NEX
|
Licensing follows a project assessment of the institution's rules, data integration, model configuration and operating scope. | Data protection terms, storage locations and hosting responsibilities are agreed for each project. The website privacy policy covers browsing and contact activity only. | Setup covers data feeds, permissions, model configuration and rule testing, scoped in the project assessment. A standard implementation fee is not listed. | Vendor Published |
T-NEX sells Fraud Detection by project quote after a demonstration, in managed operation or in a cloud account the institution holds.