Interspectral joins Vinnova-backed TRUSTAM consortium to develop federated AI for AM quality control

Interspectral, a Swedish AM data visualization software company, has joined the TRUSTAM consortium, a project funded by Sweden's national innovation agency Vinnova with multi-million Swedish krona backing. The consortium aims to develop federated AI solutions specifically for additive manufacturing quality control, enabling distributed machine learning across multiple production sites without centralizing sensitive process data. Interspectral brings its volumetric visualization and analysis platform to the project, which includes unnamed industrial partners and a defined commercialization path for the resulting process control tools.

This consortium addresses a persistent gap in AM production: quality control data remains siloed across individual machines and facilities, preventing the industry from building robust, generalizable defect detection models. Federated AI allows multiple manufacturers to collaboratively train models on their combined process data without exposing proprietary build parameters or part geometries. This directly attacks the data-sharing bottleneck that has limited AI adoption in AM quality assurance, particularly in regulated verticals like aerospace and medical where data confidentiality is paramount. The project positions Interspectral at the intersection of the software-service and cross-process segments, competing indirectly with platforms like Identify3D and Oqton that offer centralized data pipelines, but with a privacy-preserving architecture that could lower adoption barriers for risk-averse manufacturers.

For Interspectral, the practical challenge is translating consortium research into a commercially viable product that integrates with existing MES and monitoring workflows. The company must demonstrate that federated models trained across heterogeneous machine fleets (LPBF, DED, binder jetting) achieve detection accuracy comparable to centralized approaches. Buyers in aerospace and medical should evaluate whether the consortium's output includes reference implementations compatible with their existing sensor and data infrastructure, as interoperability will determine whether this remains a research project or becomes a deployable tool.