The latest articles on DEV Community by Rishi Sharma (@rishi5991). https://dev.to/rishi5991 https://media2.dev.to/dynamic/image/width=90,height=90,fit=cover,gravity=auto,format=auto/https:%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Fuser%2Fprofile_image%2F4142327%2Ffef6290e-0e25-4d80-833b-e96658788483.png https://dev.to/rishi5991 en Rishi Sharma Fri, 25 Sep 2026 05:56:34 +0000 https://dev.to/rishi5991/from-fraud-score-to-fraud-investigation-building-an-agentic-fraud-investigator-with-tigergraph-284d https://dev.to/rishi5991/from-fraud-score-to-fraud-investigation-building-an-agentic-fraud-investigator-with-tigergraph-284d <p>Fraud detection is often framed as a classification problem:<br> transaction → model → fraud probability</p> <p>That is useful, but it is not enough for an investigator.<br> A real investigation asks different questions:</p> <ul> <li>What is connected to this transaction?</li> <li>Is the same device associated with other cards?</li> <li>Has the customer or card appeared in previous investigations?</li> <li>Does the transaction belong to a suspicious temporal sequence?</li> <li>What evidence supports fraud?</li> <li>What evidence supports legitimacy?</li> <li>What information is still missing?</li> <li>Should we act now or gather more evidence?</li> <li>What action is allowed under policy?</li> <li>Can every final claim be traced back to evidence? For Hacker House Goa 2026, we built an <strong>agentic fraud investigation system</strong> using <strong>TigerGraph, GraphRAG</strong> and <strong>LangGraph</strong> to explore exactly that problem. The result is less like a fraud classifier and more like a bounded digital investigator.</li> </ul> agents ai database security