SUPPORT

Frequently asked questions

About Syntheticr

  • Syntheticr is an AML testing platform designed to replace production data as the basis for testing, training, and measuring AML system performance.

    It provides precisely engineered synthetic datasets with known ground truth, combined with objective performance scorecards. This allows teams to understand how AML systems actually perform across scenarios, segments, and money-laundering typologies, without the legal risk, access constraints, or biases of production data.

    Syntheticr is used for one-off assessments, system and vendor comparisons, and ongoing testing and improvement workflows.

  • Production data was not designed for testing or training. It is legally sensitive, difficult to access, historically biased, and a severe lacks of labels - confirmed money laundering is rare.

    As a result, testing on production data often produces weak or ambiguous evidence. Systems can appear to improve while performance gaps remain hidden, especially when using adaptive or AI-based models.

    Syntheticr replaces production data with a controlled alternative designed specifically for objective testing.

  • Syntheticr is not a generic synthetic data generator.

    Syntheticr provides purpose-built AML testing datasets with known ground truth, combined with objective performance scorecards and repeatable workflows. The focus is not data realism alone, but measurable evidence of detection performance across scenarios, segments, and typologies.

  • Syntheticr’s initial datasets are calibrated to the UK financial system, including transaction structures, customer behaviour, and economic context.

    However, the money-laundering methodologies, network typologies, and risk patterns embedded in the datasets are global in nature and reflect internationally recognised money laundering techniques.

    Syntheticr is designed to be configurable. Datasets can be adapted or custom-built to reflect different jurisdictions, institutions, products, or risk priorities, allowing teams to meet precise regional or organisational needs without relying on production data.

  • Syntheticr does not produce regulatory reports or replace governance processes.

    Instead, it provides objective, defensible evidence about AML system performance that teams can use within their existing model risk, assurance, and governance frameworks. How that evidence is presented or documented is determined by each organisation.

  • Yes. Syntheticr can be used entirely through manual uploads and one-off assessments. APIs are available for teams that want to embed testing into automated workflows, but they are not required.

  • Syntheticr is designed to complement existing AML systems, not replace them.

    Syntheticr is not:

    • A transaction monitoring or case management system

    • A regulator-facing reporting or compliance submission tool

    • A generic synthetic data generation platform

    • An “AI-first” or black-box decision engine

    Syntheticr exists to provide objective evidence of AML system performance without relying on production data. description

  • You can start a free 30-day trial of either the Starter or Pro tiers at any time here.

    If you’re interested in an Enterprise subscription, please contact us at hello@syntheticr.ai.

  • Once you activate your trial or subscription, you will receive an email from hello@syntheticr.ai containing a unique Open DeltaShare access link.

    • The link is valid for 7 days

    • It can be used multiple times during that period

    • If the link expires, you can request a new one by contacting hello@syntheticr.ai

    Alternative delivery methods are available. If you would prefer Databricks-to-Databricks DeltaShare, SFTP, or cloud-based syncing, please contact hello@syntheticr.ai.

  • Your access includes multiple related datasets (tables) representing a complete financial ecosystem, including:

    • Customer profiles (individuals and businesses)

    • Financial transactions across multiple payment types

    • Entity relationships and network structures

    • Risk intelligence-style outputs such as alerts, SARs, and exits

    The exact scope depends on your subscription tier.

  • Alongside your dataset, you will receive a Data Dictionary.

    The Data Dictionary includes:

    • Table definitions

    • Field-level descriptions and data types

    • Join logic between tables

    • Reference and lookup values

    All tables include bank_id and bank_name fields. When joining tables, always include bank_id in your join conditions to ensure accuracy and avoid mixing data across financial institutions.

  • The number of financial institutions depends on your subscription tier:

    • Starter: one financial institution

    • Pro: two financial institutions

    • Enterprise: up to six financial institutions

    Each institution is consistently represented across all tables using bank_id and bank_name.

Getting Started

  • All datasets simulate 24 months of activity.

    • Months 1–18 include full transaction history and risk intelligence-style outputs (alerts, SARs, exits)

    • Months 19–24 continue transaction activity without risk intelligence

    This structure supports both benchmarking existing systems and developing new detection approaches.

  • Risk intelligence is provided at the entity level and includes:

    • AML alerts

    • SARs filed

    • Financial crime risk exits

    As in real-world environments:

    • The data contains both true positives and false positives

    • Not all criminal entities are identified

    • True positive rates increase from alerts to exits

    Pro and Enterprise tiers also include granular transaction monitoring alert records with alert-level SAR filing flags.

  • Once you’ve downloaded your data, you should process it through the detection capabilities you want to test.

    This typically involves identifying suspected criminal entities and criminal networks based on your system’s outputs, models, or investigative workflows.

    Common approaches include:

    • Running the data through an existing AML platform

    • Applying custom detection rules or models

    • Performing investigative or analytical review of transactions and networks

    • Comparing outputs across different systems or configurations

    The goal is to identify which entities and networks you believe are involved in criminal activity.

  • Performance Baseline
    Run the data through your existing AML system to establish a clear, objective view of current performance and identify where improvements will have the greatest impact.

    Vendor Assessment
    Process the same datasets through multiple technologies to objectively compare detection performance using a consistent benchmark.

    Model Development
    Use the data to build, train, and refine detection models in a risk-free environment. Where supported, API-based workflows can be used to create automated feedback loops.

    System Optimisation
    Test changes to rules, thresholds, or models before release, then reassess performance. Where supported, API-based workflows can be used to create automated feedback loops for continuous improvement.

    Data Impact Assessment
    Quantify how changes to data scope or quality affect detection performance before making production data investments.

    System Benchmark
    Track performance over time by rerunning assessments and comparing results across multiple submissions.

Processing the Data

Submitting Results and Performance Scorecards

  • After processing the data, submit the entity IDs representing suspected criminal entities and networks.

    Requirements:

    • Supported formats: CSV, XLSX, Parquet

    • One entity ID per row

    • Entity IDs must match those in the datasets

    Important:
    Submissions containing 30% or more of the total entities may be rejected, as they do not represent a realistic detection outcome.

    Enterprise subscribers may submit results programmatically via API.

  • All scorecards include a full set of 13 performance tables, covering:

    • Overall detection performance

    • Performance by financial institution

    • Customer and entity segmentation

    • Money laundering methodologies

    • Transaction type and detection timing

    • Alert quality and prioritisation

    • Network and cross-institution network detection

    Starter (Basic) Scorecard

    Includes all 13 performance tables

    Quantitative metrics only

    Pro and Enterprise (Advanced) Scorecard

    Includes all 13 performance tables

    Added written analysis explaining:

    • Overall strengths and gaps

    • Institutional performance patterns

    • Customer and risk segmentation insights

    • Methodology coverage and blind spots

    • Transaction and alert quality observations

    • Network detection effectiveness

    • A prioritised summary of improvement opportunities

  • The response time varies by subscription tier:

    • Starter: up to 72 hours

    • Pro: up to 48 hours

    • Enterprise: up to 24 hours

    Scorecards are delivered as PDFs. Enterprise subscriptions may also receive results via API.

  • Scorecards allowances are set monthly per tier, and do not roll over:

    • Starter: 1 Basic Scorecard per month

    • Pro: 3 Advanced Scorecards per month

    • Enterprise: Custom quantity per month

  • Yes. Users are expected to work iteratively on the same dataset, refining detection approaches and resubmitting results.

    Once you reach a high detection threshold, you will be granted access to a more challenging dataset.

  • If you achieve a detection rate above 80%, you will be notified and granted access to a larger, more challenging dataset.

  • All subscription and billing management is handled through the customer portal, including plan changes, billing details, and cancellations.

    You can access the customer portal at any time using the secure link provided in your subscription and billing emails from hello@syntheticr.ai.

  • You can request an upgrade at any time by contacting hello@syntheticr.ai. Any changes take effect at the end of the current billing period unless otherwise agreed.

    You can access the customer portal at any time using the secure link provided in your subscription and billing emails from hello@syntheticr.ai.

  • Cancellations are managed through the customer portal and take effect at the end of the current subscription period.

    After cancellation, you will be asked to confirm deletion of all Syntheticr datasets, which are uniquely watermarked.

    You can access the customer portal at any time using the secure link provided in your subscription and billing emails from hello@syntheticr.ai.

Account Management