Years in the making. Unrivalled in performance.
ULTRA is Ripjar's proprietary intelligence engine and the foundation of everything on the Ripjar platform. It reads 10bn+ historical articles and 5m+ new ones each day across 150+ languages, and resolves them into roughly 15m entity profiles at 98.7% accuracy. The result is the system of record for entity risk that Tier 1 banks, governments and Global Fortune 500 enterprises rely on.
ULTRA was the British codename for the intelligence produced from decrypted Axis signals at Bletchley Park, the forerunner of GCHQ, where Ripjar's founders worked. The engine takes its name from that intelligence.
ULTRA is the engine that lets compliance teams defend every screening decision, with the accuracy, coverage and audit trail to prove it.
ULTRA is a governed, tested and reproducible AI system, built for financial crime compliance and running in production at Tier 1 banks today.
ULTRA is a framework of proprietary transformer models, graph techniques, specialised machine learning and generative AI. Each is applied where it is best suited to the task.
ULTRA reads 5m+ new articles each day across 150+ languages. Around half of the articles ULTRA reads are duplicate or noise, and machine learning and plagiarism detection strip them out before they cost an analyst any time.
In a single article a name can appear as perpetrator, lawyer, journalist or victim. ULTRA works out which is which. Proprietary graph algorithms and custom AI resolve every mention to the correct profile, and sanctions, PEPs, other watchlists and adverse media merge into one Dynamic Risk Profile per entity.
Transformer models attribute risk to the entity that carries it and rank every match on a confidence scale, so sanctions evasion surfaces while irrelevant matches on common names clear. Once your team dismisses a link, ULTRA stops surfacing it. 99% fewer alerts reach an analyst.
Every article ULTRA reads, every query it answers and every decision your analysts make writes back to a resolved profile, so the next query is answered from what is already known.
Historical data is re-run against new models, so an analyst onboarding a new client can surface risk in an old article.
Ripjar has been building this record for years. It is why screening on ULTRA gets faster and more accurate the longer it runs.
articles in the system of record
resolved entity profiles
years of analyst reading in a single day's intake
Created by former GCHQ technologists for environments where security, privacy and trust are conditions of entry.
An MCP interface makes ULTRA callable from the agentic frameworks you already run. That includes Screening Assistant, Ripjar’s own agentic AI, which is also delivered inside the platform.
Deploy as multi-tenant or as a logically separated single-tenant instance, in the regions you operate in. Your data stays under your policies and your privileges, wherever it runs.
Data visibility is restricted to those entitled to see it, with privacy and security by design. ISO 27001 certified. SOC 2 compliant.
Every decision, alert and summary is backed by evidence, traceable to source and held in an authoritative, reproducible system of record. A regulator asking what you knew on a given date gets a fast, defensible answer.
ULTRA is Ripjar's proprietary intelligence engine and the foundation of everything on the Ripjar platform. It reads 10bn+ historical articles and 5m+ new articles each day across 150+ languages, resolves them into roughly 15m entity profiles at 98.7% entity accuracy, and maintains the system of record for entity risk. Ripjar is trusted by 400+ customers worldwide, including 25% of Global Systemically Important Banks and 35+ Global Fortune 500 enterprises.
A general-purpose model can read text at scale. ULTRA adds three things built for financial crime compliance. A framework of specialised models resolves entities and attributes risk. A system of record holds a resolved profile per entity and answers each query from what is already known. And every output carries an audit trail back to its source articles.
ULTRA powers all of Ripjar's products, including Ripjar Screening, Screening Assistant and Ripjar Labyrinth. It also powers all their capabilities, including Watchlist Screening, Adverse Media Screening and Dynamic Risk Profiles.
Yes. A Model Context Protocol (MCP) interface makes ULTRA callable from the agentic frameworks an organisation already runs, including Screening Assistant, Ripjar's own agentic AI.
Every alert, decision and summary ULTRA produces is backed by evidence and traceable to its source. Summaries link to the exact source articles and confidence scores show how certain each match is, so an auditor can see what was known on a given date and why a decision was made.
A Ripjar specialist will walk you through ULTRA on the sources and thresholds your team screens against every day, and show what the record already holds on the names carrying the most risk.