Ripjar has introduced new AI capabilities in ULTRA, covering alert triage, entity matching, and adverse media risk classification.
Ripjar has updated ULTRA, its screening engine for financial institutions and enterprises, with capabilities focused on alert triage, entity matching, and adverse media classification. The updates are intended to support compliance teams in identifying and assessing risk.
ULTRA combines data for risk screening
ULTRA uses proprietary graph algorithms, transformer models and AI to process information across more than 150 languages. Ripjar said the platform has resolved more than 10 billion historical articles and processes over 5 million new articles each day. The engine combines data from politically exposed persons (PEPs), sanctions, watchlists, and adverse media into a single entity profile. Ripjar said ULTRA achieves 98.7% accuracy in entity resolution and reduces the volume of alerts reaching analysts by 99%. Resolved entities are retained, allowing previous decisions to be used for future screening activity.
The latest updates expand Screening Assistant, Ripjar’s agentic AI capability for alert triage. The changes are designed to reduce routine alert handling and support compliance teams in assessing potential risks. ULTRA’s entity-matching capabilities have also been updated to distinguish between parent organisations, subsidiaries, and similarly named entities within corporate groups. This is intended to help determine whether an alert relates to the specific entity being screened. The update expands Screening Assistant’s handling of complex naming conventions, including Arabic names.
Adverse media classification receives AI upgrade
Ripjar upgraded ULTRA’s adverse media risk classifiers, which assess news articles to determine whether they relate to a specific risk and identify the person involved. The model upgrades use machine learning for classification and entity-resolution tasks. ULTRA is designed for environments where screening decisions need to be explainable and auditable. Alerts, decisions and summaries are supported by evidence that can be traced to their source and retained in an auditable system of record.
The platform uses multiple AI layers to remove duplicate information, resolve mentions into entity profiles and associate risks with the relevant person, according to Chief Product Officer Conrad Nicholas. He added that ULTRA combines these models while maintaining explainability.
CEO Matt Mills said changing sanctions, increasing adverse media and more sophisticated criminal activity are challenges for compliance teams. He also highlighted the need for visibility into how AI-supported screening decisions are reached.
