Detecting corporate manipulation before the market reacts.
An AI-powered forensic intelligence platform that analyzes NSE-listed companies using quantitative fraud detection models, anomaly scoring, peer-relative analysis, and LLM-generated investigative narratives.
AuditGPT is a forensic financial intelligence platform engineered to identify potential accounting manipulation, distress indicators, and anomalous corporate behavior across Indian equity markets.
Built during the IAR Udaan Hackathon 2026, the platform processes over a decade of NSE financial data and evaluates companies using multiple forensic accounting frameworks simultaneously — including Beneish M-Score analysis, Altman Z-Score modeling, industry-adjusted anomaly detection, and structural trend-break analysis.
Unlike traditional screeners that rely on static thresholds, AuditGPT benchmarks every company relative to sector-wide peer behavior, enabling more context-aware risk evaluation and anomaly identification.
The system generates a unified fraud-risk score ranging from 0–100 and visualizes:
financial manipulation indicators
distress probability
peer-relative deviations
red-flag timelines
sector-wide contagion heatmaps
AI-generated forensic narratives
The platform also includes a dedicated historical reconstruction of the Satyam scandal, replaying year-by-year financial deterioration and manipulation signals leading up to one of India’s largest accounting fraud cases.
The frontend was developed using React 19, Vite, React Router, and Tailwind CSS with a Bloomberg-terminal-inspired visual system optimized for dense analytical workflows and high-information dashboards.
The backend was built with Express.js and TypeScript, serving forensic scoring pipelines, search endpoints, sector intelligence APIs, and report-generation systems. Google Gemini 1.5 Pro was integrated to synthesize structured financial findings into human-readable investigative summaries and narrative risk reports.
Core analytical systems include:
Beneish M-Score fraud detection
Altman Z-Score bankruptcy analysis
Industry-relative z-score anomaly detection
Trend-break and structural variance analysis
Sector-level threat mapping
Financial sentiment aggregation
Company risk clustering
Additional functionality includes:
live fraud radar dashboards
searchable NSE company intelligence
Telegram bot integration for mobile queries
sector comparison engines
narrative stream APIs
critical-risk heatmaps
AuditGPT was designed as a production-grade financial intelligence concept combining quantitative finance, AI reasoning systems, and investigative visualization into a single unified platform experience.
