Developing an AI-driven, integration-ready workflow service

Overview

Industry

Location

Software and IT services

Belgium

Technology Used

Azure Container Apps

Azure OpenAI Service

Azure SQL Database

Azure Storage Accounts / Blob Storage

Docker

Figma

Python

Qdrant

REST API

Project overview

To create an AI solution to handle the process of identification of products based on photos, descriptions and material data before moving them further into ERP and PIM systems.

Client background

A software development and consulting company focused on digitalization, innovation and software solutions that help clients optimize business processes, with particular experience in supply chain-related environments.

Goals

Our client wanted an AI-driven solution to support the path from a raw supplier input to structured, compliance-oriented classification results. Our team was tasked with delivering an AI-based workflow for product intake and classification, covering taxonomy, color, HS code and EUDR checks. Our goal was to leverage AI to accelerate product identification and reduce the amount of manual work needed to prepare product data for further processing.

Results

High-level of accuracy

The combination of solid engineering and good-quality reference data led to overall accuracy up to 90%.

Faster intake process

Integrating secure AI-driven automation sped up workflows and reduced repetitive manual effort.

Practical functionality

Delivering a solution to classify single products and full vendor datasets, result storage, manual correction support and export-oriented outputs for downstream use.