Overview

GB-FiscalAI is an artificial intelligence-based fiscal anomaly detection system, developed for Guinea-Bissau's customs services.

Objectives

• Automate the analysis of customs declarations
• Detect anomalies and potential fraud
• Prioritize controls according to risk level
• Optimize the efficiency of customs agents

Main Features

Risk Analysis

The AI engine analyzes each declaration according to several criteria:

Criteria	Description
Undervaluation	Comparison with reference prices
History	Importer's past behavior
Country of origin	Risk associated with source country
HS Code	Sensitivity of declared product

Risk Levels

Low (0-30%)

Declaration compliant, random inspection possible

Medium (31-50%)

Documentary verification recommended

High (51-75%)

Physical inspection required

Critical (76-100%)

In-depth inspection mandatory + Alert generated

Architecture

┌─────────────────────────────────────────┐
│              Web Interface               │
│         (React + TypeScript)            │
├─────────────────────────────────────────┤
│              REST API                    │
│             (FastAPI)                   │
├─────────────────────────────────────────┤
│    ML Engine    │    Database             │
│  (scikit-learn) │    (PostgreSQL)       │
└─────────────────────────────────────────┘

Prerequisites

• Modern web browser (Chrome, Firefox, Edge, Safari)
• Stable internet connection
• Active user account

Support

For any assistance, contact the technical team of the Ministry of Finance.