ATB Joins euroFMX to Advance European Manufacturing AI

ATB Bremen has joined euroFMX, a major European research and innovation initiative developing trustworthy generative and agentic artificial intelligence for manufacturing.

Manufacturing companies are increasingly exploring generative AI for engineering, production planning, quality management and operational support. However, general-purpose AI models are not designed to interpret the physical constraints, technical terminology and process dependencies found in industrial environments. They may also lack the traceability, data protection and reliability required when their outputs affect production systems.

euroFMX addresses this gap by developing an open European technology stack for industrial generative AI. The project will combine manufacturing data, specialised AI models, industrial agents and sovereign computing infrastructure to support the safe use of AI in engineering and production.

The four-year initiative brings together 72 partners from 20 European countries and has a budget of €45 million. A further €10 million is reserved for open calls intended to involve additional companies, developers and technology providers.

AI designed for industrial conditions

The central objective of euroFMX is to develop AI models that understand the conditions under which manufacturing systems operate.

Industrial AI must be able to work with heterogeneous information from machines, sensors, engineering systems, quality records and production-management platforms. It must also account for physical relationships, safety requirements, process history and operational constraints.

euroFMX will therefore develop an end-to-end industrial AI stack covering:

• high-performance and sovereign computing infrastructure;
• secure industrial data products and data spaces;
• manufacturing-specific generative and frontier models;
• orchestration of industrial AI agents;
• validation and scaling in operational manufacturing environments.

The planned models will combine multimodal reasoning with industrial semantics, long-term process information and physics-informed methods. Industrial agents will use these models to interact with operators, software tools, machines and production platforms.

A key requirement is that these systems remain trustworthy and human-centred. AI should support industrial specialists and improve their ability to respond to complex situations. It should not introduce decisions that cannot be understood, supervised or corrected by the people responsible for the production process.

Validation in ten manufacturing pilots

The euroFMX technologies will be evaluated through ten industrial pilots covering machinery, manufacturing processes and supply chains.

The pilots include applications in battery remanufacturing, robotic welding, dimensional metrology, railway refurbishment, automotive electronics, aluminium recycling, medical-product manufacturing, aerospace additive manufacturing, electric-powertrain production and laser-cutting systems.

These demonstrators will investigate how industrial AI can support objectives such as:

• flexible reconfiguration of production systems;
• earlier detection of quality problems;
• reduced production interruptions;
• more efficient use of materials and energy;
• improved planning under changing conditions;
• closer cooperation between operators and AI-supported systems.

The pilots are intended to test the technologies under realistic industrial constraints rather than only in isolated laboratory environments.

ATB’s contribution

ATB contributes its experience in industrial software architectures, context-aware systems, semantic interoperability and human-centred AI.

Industrial AI agents need access to more than individual sensor values or documents. They must understand the current task, the state of the production system, the role of the user and the relationships between information from different technical systems.

ATB’s work in European and national research projects has addressed these challenges through context modelling, knowledge representation, interoperable software services and the integration of AI technologies into operational environments.

Within euroFMX, ATB will help connect manufacturing AI with the processes, information sources and software systems in which it must operate. Relevant technical questions include how agents receive reliable contextual information, how different AI and software components exchange knowledge, and how responsibility is shared between automated systems and human experts.

This work builds on ATB’s research in projects including AI4Work, SOPRANO, ASTIR, XR5.0 and GeniVerse. These projects investigate human–AI collaboration, adaptive industrial assistance, intelligent robotics software and the coordination of work between people, AI agents and machines.

Strengthening European industrial AI

euroFMX is designed as an open European initiative. Its results are intended to support manufacturers, technology providers, research organisations and small and medium-sized enterprises in adopting industrial AI without becoming dependent on closed, general-purpose platforms.

The project will also address skills, standards, regulatory requirements and the wider ecosystem needed to transfer industrial generative AI into practice.

For ATB, participation in euroFMX provides an opportunity to apply its software and integration expertise within one of Europe’s largest manufacturing AI initiatives. The objective is not simply to introduce more AI into factories, but to develop systems that are technically reliable, industrially useful and compatible with European requirements for human oversight and data sovereignty.

ATB welcomes contact from manufacturing companies and research organisations interested in industrial AI, context-aware systems, semantic interoperability and human–AI collaboration.

Project facts

• Project: euroFMX – Foundational Models for Trustworthy Industrial Generative AI in Europe
• Duration: 48 months
• Consortium: 72 partners from 20 countries
• Budget: €45 million
• Open-call funding: €10 million
• Industrial pilots: 10
• Grant agreement: No. 101299128