FraudShield Release v1.6.0
This release introduced the AI Engine, adding AI model training, AI risk scoring for every processed transaction and auto case creation based on the AI risk score, and split API keys into Active and Expired tabs.
The AI Engine reaches across many existing pages, so its changes are listed first under each type in the table below. To train and enable your first model, see Training and Activating a Model. To learn what the score means and how to use it alongside your rules, see How AI Risk Scoring Works.
| Type | Feature | Description |
|---|---|---|
| New | AI Models | Trains AI models on Finalised rule plans and tracks the progress and result of every training pipeline. |
| New | AI Risk Scoring | Scores each processed transaction from 0 to 100 using the model enabled on the active rule plan, independently of rule results. |
| New | Auto Case Creation Based on AI Risk Score | Creates a case with the matching risk level when a transaction triggers no rules but its AI risk score falls within an enabled range. |
| Improvement | Rule Plans | Lets you train AI models on Finalised plans and enable one trained model each for Production and Test. |
| Improvement | AI Risk Assessment | Transaction and case details show the AI risk score with a rule breakdown of Impact, Influence and Direction. |
| Improvement | Cases | Adds a sortable AI Risk Score column, Min AI Risk Score and Max AI Risk Score filters, and an AI Risk tile in case details. |
| Improvement | Customer Cases | Adds a sortable AI Risk Score column to the customer Cases tab. |
| Improvement | Case Reports | Exported case PDFs include the AI Risk Assessment and the AI values for each processed rule. |
| Improvement | Batch Upload | Shows the ID of the AI model pipeline that scored each batch. |
| Improvement | Audit Logs | Records AI model training, pipeline status changes and model activation, and adds a Revert Rule entry when a default rule is reverted. |
| Improvement | User Roles | Gives Rules Manager full access to AI Models, including training and enabling models, and Rules Viewer read-only access. |
| Improvement | API Keys | Separates keys into Active and Expired tabs, showing when each expired key lapsed. |
| Change | Submit Single Transaction API | The response now includes an aiRiskAssessment object and per-rule influence and impact values. |
| Change | Login Page | Uses a refreshed design. |
This release also includes minor performance improvements, bug fixes and internal updates.