Enrichment and Research in Greenlite AI
Last updated: June 23, 2025
Overview
Greenlite has a proprietary research pipeline it utilizes to find more information on match entities and customer entities. The system automatically enriches profiles with relevant external data to provide comprehensive context for alert investigations.
How the Research Pipeline Works
Initial Data Collection: Greenlite takes in all information from the initial alert input and customer data input
Profile Creation: Uses this information to create a profile on the customer and match entity respectively.
OSINT Research: Conducts OSINT (Open Source Intelligence) across all reputable sources on the web
Relevancy Scoring: Assigns a relevancy score to each source and only surfaces the most relevant in the investigation
Profile Enrichment: The investigation enriches the profile with relevant findings
Documentation: Greenlite keeps detailed AI notes of all sources it surfaces as well as notes on the updates made to the profile
Research Relevancy Scoring
OSINT Relevance Scoring (1-5 scale):
1. Irrelevant or Misleading: Incorrect, outdated, or unrelated information - filtered out by default
2. Low Relevance: Adds little value, only confirming well-established facts
3. Moderately Relevant: Contains useful information supporting known facts or introducing tangentially related information
4. Highly Relevant: Provides significant, directly applicable information to the investigation
5. Critical Source: Contains crucial information that could significantly impact or redirect the investigation
Greenlite typically uses relevancy levels 4 or 5 in its investigations to ensure only the most accurate findings populate profiles.
Governance and Quality Assurance
Greenlite's research pipeline is extensively tested and vetted. Greenlite is consistently testing its pipeline to ensure quality and maintain the highest standards of accuracy and reliability for compliance investigations.
Configuration Options
Organizations can configure:
Relevancy thresholds for different types of investigations
Which sources to prioritize or exclude
How findings are incorporated into profiles