Documentary Fraud Analyst
Purpose
Documentary Fraud Analyst performs documentary fraud assessment on documents such as text, invoices, PDFs, and images. Its purpose is to extract relevant facts, run forgery and inconsistency checks, score fraud risk, and produce an evidence-backed checklist of next actions.
Primary users
The primary users are external client-facing insurance and reinsurance stakeholders who need support assessing potentially fraudulent or inconsistent documentation. The agent is owned by Hugo RUIZ.
Where it fits (process/stage/trigger)
The agent fits into an insurance and reinsurance document review process when document references, contextual information, and optional attachments need to be assessed for fraud indicators. It is triggered when a user provides document context in plain text and, when available, PDF or image attachments.
Key capabilities / workflow
The agent extracts facts from the provided document references and context, reviews available document payloads, performs forgery and inconsistency checks, identifies contradictions and fraud signals, scores fraud risk, and generates an evidence-backed report with next actions. It can use MCP-connected documentation and tool outputs in the mock version, along with extracted document payloads.
Inputs
Inputs include document references and context in plain text, with optional PDF or image attachments. The available dataset includes MCP-connected documentation and tool outputs in the mock version, as well as extracted document payloads.
Outputs / Deliverables
Outputs include an evidence-backed report containing an extracted facts table, fraud signals, a fraud risk score, contradictions, recommended next actions, and missing data.
Value
Documentary Fraud Analyst helps client-facing insurance and reinsurance users structure document fraud assessment, highlight evidence-backed inconsistencies, and identify next actions. It supports faster, more consistent review of suspicious documentation while clearly surfacing missing data and risk indicators.
