CartographAI
Accelerating AML Data Integration
How CartographAI demonstrated a step-change in AML data integration, from months to days.
CartographAI, an end-to-end data integration tool for transaction monitoring systems, set out to prove that data integration for financial crime systems could be done in a fraction of the time it currently takes.
A new transaction monitoring implementation routinely takes 12-24 months, with much of that time lost to the slow, manual work of mapping source data into an application's target schema. CartographAI automates that work by combining subject matter expertise with AI to analyse source data, generate mapping rules, and convert them into executable code. To show it worked at the scale and complexity of a real deployment, CartographAI needed realistic data and a demanding target to map it to. NICE Actimize, one of the most widely deployed transaction monitoring platforms provided the target and Syntheticr provided the data.
Syntheticr datasets are delivered in a normalised schema, which is ideal for generating well-structured financial transaction data, but that may be difficult for some AML systems to ingest. Actimize is a good example, as it ingests data in its own denormalised, application-specific schema. Bridging that gap is precisely the problem CartographAI exists to solve, and Syntheticr data provided an ideal proving ground. By mapping the Syntheticr data into the Actimize schema, CartographAI developed, tested, and demonstrated its platform without requiring access to sensitive production data.
Testing a data mapping platform is only useful if the data used replicates what is typically encountered in a real-world implementation. Specifically, the data needed the scale of a large financial institution, the full breadth of entities, transactions, and behaviours, and the nuanced, low-frequency typologies where data quality and structure affect outcomes. Syntheticr provided exactly that. CartographAI applied its platform to the Syntheticr data, analysing the normalised schema, mapping it field by field to the Actimise denormalised schema, and generating the SQL needed to perform the transformation.
This delivered three clear benefits for CartographAI:
Production-Scale
Using the Syntheticr data, CartographAI tested its application at a scale several orders of magnitude greater than what the team could have built internally. This exercised the mapping engine beyond the Production-scale volumes expected in a large financial institution and surfaced field-level and structural challenges that normally appear only in production.
Edge-Case Validation
The breadth and fidelity of the Syntheticr data model meant the platform was exercised not only against core AML use cases but also against more nuanced, edge-case typologies, where field population and structure most affect outcomes. This provided a far more honest measure of its capability than a simpler dataset would have.
Product Improvements
Processing the Syntheticr data through the CartographAI platform highlighted several opportunities to enhance the product's core functionality, which will feed directly into the roadmap for future versions.
For Syntheticr, the work also produced outcomes that have been used to improve the product:
1
CartographAI's mapping and the resulting SQL now allow Syntheticr datasets to be delivered directly in an Actimize schema. As a result,
organisations using Actimize can load high-fidelity synthetic data into their transaction monitoring systems and immediately start testing performance, rather than spending time on data integration.
2
Mapping the Syntheticr model against a demanding real-world schema surfaced specific opportunities to refine the underlying data
structure. These improvements will further refine the Syntheticr datasets, ensuring they remain the class-leading mechanism for testing transaction monitoring models and precisely measuring what they detect and what they miss.
3
The SQL transformation is now built into Syntheticr’s data generators, allowing custom Syntheticr datasets to be produced natively in the
Actimize schema on demand. Customers can already shape the data according to the specific scenarios they want to test, but now they can also specify that the data be provided in an Actimize schema
The overall outcome was highly valuable for both CartographAI and Syntheticr resulting in meaningful improvements to both products that will directly benefit their customers.
CartographAI validated its platform against a realistic, production-scale environment, proved it could map a complex source ecosystem into the Actimize schema with confidence, and identified concrete improvements for future client engagements.
Syntheticr gained a faster, friction-free path for Actimize customers, a higher-quality dataset, and the ability to deliver bespoke data in a commonly used schema that is ready-to-integrate.
For their customers, this combination significantly eases the implementation, testing, and tuning of transaction monitoring systems, eliminating the months of expensive data integration work often involved.
Experience how easy it can be to test and train your AML system
Start a free trial to get an objective performance insight without production data.