Turn unstructured bank statement PDFs into a graph for investigating suspicious transaction patterns. See how AI-powered data extraction and PuppyGraph help you trace money across accounts, uncover shared counterparties, and explore circular transaction paths.This demo walks through a pipeline using PyPDF to extract text, GPT-4o to structure account and transaction details, and CocoIndex to manage incremental processing into PostgreSQL. PuppyGraph then queries those PostgreSQL tables directly as a graph, with no additional ETL into a separate graph database.Follow along to:

• Connect PostgreSQL to PuppyGraph and map your data to nodes and edges.
• Explore accounts, statements, and transaction relationships in a visual graph.
• Run a graph query to find accounts connected through a shared counterparty.
• Ask the built-in AI assistant to investigate money returning to an account through multiple intermediaries.

Whether you’re exploring bank statement analysis, financial investigations, or graph analytics, this demo shows how to move from PDF documents to connected data you can query in plain English.