The primary authorised push payment (APP) fraud solutions that integrate with existing payment systems are Vyntra, Featurespace, Feedzai, BioCatch, and NICE Actimize.
These platforms provide real-time transaction analysis, artificial intelligence (AI) risk scoring, and automated investigation tools designed to connect directly with core banking infrastructure, payment rails, and digital channels. This article compares how each platform handles integration with existing systems and legacy software.
APP fraud solutions that integrate with existing payment systems
Overview of APP fraud solutions and integration methods
Platform | Integration approach | Key features |
Vyntra | Sits on top of the existing system, integrating risk scoring models within legacy payment processing flows for cloud or on-premises | 100 millisecond risk scoring, pre-built artificial intelligence models, and a unified case management workspace |
Featurespace | Integrates through real-time APIs and data models that embed its flagship ARIC™ Risk Hub | Adaptive behavioural profiling, Cambridge-developed analytics, and false positive reduction |
Feedzai | APIs to the RESTful interface with cloud-based connectors | Unified RiskOps workspace, explainable artificial intelligence scoring, and live session telemetry |
BioCatch | Deploys software development kits to frontends while feeding scores into backend risk engines | Behavioural biometrics tracking, physical interaction metrics, and passive session monitoring |
NICE Actimize | Connects to legacy software and enterprise service buses using standardised connectors and batch feeds | Cross-channel data correlation, automated case management workflows, and pre-built compliance reporting |
Vyntra
Since we’re the ones writing the article, it makes sense to start with ourselves.
Vyntra provides a purpose-built financial crime platform designed to sit on top of a bank’s existing processing software without requiring changes to its core systems. The platform ingests copies of existing payment messages across various channels to analyse transaction context, user habits, and beneficiary risk in real time.
Key capabilities include:
- Deployment without system overhauls: Ingests live data streams from core software using lightweight, non-intrusive connectors to eliminate complex integration engineering.
- Real-time scam prevention: Evaluates incoming and outgoing payments using pre-built machine learning models to detect social engineering, account takeovers, and money mule patterns.
- Unified risk operations: Consolidates alert investigations into a single environment featuring natural language summaries, visual data lineage, and complete audit trails.
- Near-instant processing speeds: Scores transactions in 100 milliseconds to prevent authorised push payment fraud without causing payment timeouts or latency breaches.
Read more: How banks can stop payment scams without adding latency.
“We take more of an approach where we say don’t change anything to your systems because those systems are highly critical and highly sensitive, and so you don’t want to touch them. But in any case, those systems are already sending messages to each other. Just send us a copy of the things that you are already exchanging, and we will adapt to the messages that are already circulating.” Joris Lochy, Product Manager at Vyntra
Vyntra may not suit every bank, so here are four other established platforms worth considering.
Featurespace
Featurespace offers the ARIC Risk Hub, an enterprise fraud detection engine built on adaptive behavioural analytics. Developed from research at the University of Cambridge, the platform creates individual profiles for account holders to detect subtle changes in payment activity.
Featurespace connects to existing banking systems via application programming interfaces (APIs) and message queues. The platform analyses transaction velocity, device signals, and beneficiary history to catch APP scams as they occur.
Core capabilities:
- Adaptive behavioural profiling: Learns individual customer habits to spot anomalies caused by social engineering.
- False positive reduction: Cuts genuine transaction blocks by up to 75% compared to static rule engines.
- Flexible deployment options: Operates as a cloud-native software-as-a-service (SaaS) or on-premises installation.
Feedzai
Feedzai provides an AI-first fraud prevention architecture centred around its RiskOps Studio environment. The system ingests fragmented signals across digital banking channels, card networks, and account-to-account rails to deliver unified risk decisions.
Feedzai integrates with bank payment infrastructure through real-time event streaming and REST APIs. By combining device telemetry with behavioural analysis, the platform detects signs of coercion, active phone call manipulation, and unusual transfer sequences before funds leave the account.
Core capabilities:
- Unified RiskOps workspace: Consolidates rule configuration, alert review, and regulatory reporting into one interface.
- Explainable model scoring: Delivers clear decision rationale alongside every risk score to satisfy compliance audits.
- Device and session intelligence: Tracks session hesitation and active call indicators to detect manipulation during live payment entry.
BioCatch
BioCatch approaches APP fraud prevention through behavioural biometrics, analysing how users interact with online banking applications and mobile screens. The platform integrates into digital banking frontends via software development kits (SDKs) while feeding risk scores into backend fraud management engines.
The BioCatch Connect platform evaluates physical movements such as swipe patterns, typing cadence, device orientation, and screen pressure. When a fraudster coaches a victim through an authorised payment, BioCatch identifies physical indicators of hesitation, distraction, and cognitive overload.
Core capabilities:
- Behavioural biometrics tracking: Measures physical interaction metrics to spot remote access tools and coerced user behaviour.
- Session manipulation flags: Identifies abnormal hesitation and extended payment entry times typical of phone-based scams.
- Seamless user experience: Monitors customer interaction passively without inserting security prompts for genuine payments.
NICE Actimize
NICE Actimize delivers enterprise-scale fraud management through its ActOne investigation and surveillance platform. Built for large financial institutions, the software correlates transaction data, device intelligence, and customer records across multiple operational units.
The solution connects to legacy core banking software and enterprise service buses through standardised connectors and batch processing feeds. It combines machine learning detection with customizable business rules to address authorised push payment scams, account opening fraud, and internal threats.
Core capabilities:
- Correlated data intelligence: Blends data streams from multiple channels into a single risk profile.
- Enterprise workflow automation: Automates alert routing and case assignments across global compliance teams.
- Regulatory reporting framework: Includes pre-built templates for filing suspicious activity reports with regulatory authorities.
How APP fraud solutions integrate with existing payment systems
Connecting an anti-fraud layer to existing core infrastructure is often the most challenging phase of software deployment. Core banking software is frequently decades old, fragile, and sensitive to performance disruptions.
Modern real-time payment fraud prevention solutions avoid touching or modifying core application code directly. Instead, these platforms connect through lightweight APIs, passive messaging feeds, and pre-configured risk models that evaluate payments in parallel.
API integrations
API integrations allow fraud detection engines to exchange data with external banking modules through standard protocols. When a customer submits an authorised transfer through mobile or web channels, the digital banking system sends an API request containing transaction parameters, beneficiary details, and session metadata to the fraud tool.
Pre-built AI models
Pre-built AI models remove the requirement for financial institutions to train machine learning algorithms from scratch using raw data. These risk engines ship with pre-configured detection algorithms trained on vast datasets of payment scams, account takeovers, and money mule behaviours.
During integration, the software maps incoming payment attributes to existing feature libraries. This allows banks to go live within weeks rather than spending months on custom data science engineering.
Adaptation to existing payment flow systems
The safest way to integrate fraud technology into sensitive environments is by observing communication that already occurs between internal modules. Instead of requiring core applications to build new outbound calls, non-intrusive fraud platforms take a duplicate copy of messages already circulating across the network.
By ingesting duplicate payment traffic, the platform normalises different standards, such as SWIFT MT, ISO 20022, and regional instant schemes, inside its own processing layer without disrupting payment execution.
Available as SaaS or an on-premises add-on
Different institutions operate under distinct data sovereignty, compliance, and cloud architecture mandates. Flexible APP fraud platforms deploy as cloud-native SaaS instances or as on-premises add-ons within a bank’s private data centre. SaaS deployments provide rapid scaling, automatic model updates, and lower infrastructure overhead. On-premises deployments keep sensitive customer data entirely within the bank’s internal network perimeter.
Adds no latency to the existing process
Instant payment rails enforce strict settlement windows, often requiring transactions to clear within seconds. Adding an anti-fraud check inside the execution path must not cause payment timeouts or service-level agreement breaches. Integration layers achieve zero added latency by running risk evaluations in parallel with payment routing or by utilising dedicated, high-speed scoring endpoints capable of responding in under fifty milliseconds.
Choose a solution that integrates beyond the backend with a case manager and analytical dashboards
Backend data integration solves only half the challenge if fraud teams must still navigate disconnected software during an investigation. Linking real-time scoring engines directly to a centralised case management system provides natural language summaries and visual data lineage, allowing analysts to confirm push payment scams in minutes.
FAQs
How do APP fraud tools connect to legacy core banking software?
Most APP fraud tools connect to legacy systems by ingesting copies of transaction traffic from existing message buses or middleware without requiring edits to core code.
Will adding an APP fraud detection platform slow down instant payment processing?
No, modern fraud platforms process transactions in parallel using dedicated scoring endpoints that return decisions in milliseconds, preventing processing timeouts.
What data sources are required to detect authorised push payment scams?
Effective scam detection requires payment details, beneficiary account history, device telemetry, session interaction data, and historical customer habits.
Sources:
- https://www.featurespace.com/products/aric-risk-hub/
- https://www.featurespace.com/newsroom/featurespace-launches-aric-scam-detect-to-protect-the-financial-services-industry-and-outsmart-scammers
- https://www.featurespace.com/newsroom/natwest-improves-scam-detection-rate-by-135-using-featurespaces-technology
- https://www.about-fraud.com/providers/featurespace/
- https://www.feedzai.com/riskops/
- https://www.feedzai.com/solutions/scam-prevention/
- https://www.feedzai.com/solutions/ai/
- https://www.biocatch.com/biocatch-connect
- https://www.biocatch.com/product-descriptions
- https://www.biocatch.com/blog/authorized-push-payment-fraud-how-digital-behavior-can-uncover-social-engineering-scams
- https://www.biocatch.com/press-release/social-engineering-press-release
- https://www.biocatch.com/blog/what-is-behavioral-biometrics
- https://www.niceactimize.com/glossary/authorized-push-payment
- https://www.niceactimize.com/enterprise-risk-case-management/enterprise-risk-case-management-overview
- https://www.niceactimize.com/Documents/ActOne_Prospect_Brochure.pdf
- https://www.niceactimize.com/lps/real-time-payments-real-time-fraud
- https://www.swift.com/standards/iso-20022/iso-20022-faqs/mt-iso-20022-conversion
