BFA – Business Factory Auto

MATIAS

Contact details:

Company: XION AUTOMATIZACIONES INTELIGENTES S.L.

Contact: África Quiñones

Telephone: +34 624 288 940

Email: hola@xion-ai.es

Web: www.xion-ai.es

LinkedIn: www.linkedin.com/company/xion-ai-es/

Project description:

MATIAS is an artificial intelligence solution for the automatic estimation of CNC machining times and costs, designed for workshops and industrial suppliers in the automotive and precision manufacturing sectors. It addresses a very specific bottleneck: the quotation of complex parts, which in medium-sized workshops can require between 2 and 8 hours, depend heavily on accumulated technical expertise, and generate errors that directly impact profitability and competitiveness. In the Galician context, where a significant part of the Stellantis supply chain depends on small and medium-sized machining workshops, MATIAS aims to accelerate commercial response times and improve scalability.

Solution & technologies:

The tool interprets technical drawings in PDF and STEP files, automatically extracts geometric features and tolerances through B-Rep analysis, and generates in less than 90 seconds a detailed estimate broken down by machining process — milling, turning, grinding, threading, drilling, or EDM — including the recommended machine, operator, and total cost. It uses machine learning models trained with synthetic data generated from each workshop’s own parameters: machines, operators, materials, pricing rates, and production conditions. The architecture is deployed in a private Docker instance for each client, ensuring that quotation data and competitive know-how remain within the workshop’s environment.

Innovation contributed:

MATIAS innovates by combining industrial artificial intelligence, geometric analysis of technical files, and privacy-by-design principles in a critical process for automotive suppliers. Its key differentiator is not only accelerating quotations but also personalizing the model for each workshop from day one through proprietary synthetic data and progressive learning from real quotations. This reduces dependence on expert profiles, improves accuracy, keeps data under client control, and transforms a slow manual process into a scalable competitive advantage.