AI Talentum and AZTI are advancing Xpectra, an AI-powered quality-control platform that uses NIR spectroscopy to monitor food production in real time—without destroying samples. Its first application, fish-oil quality control, suggests it could help companies detect deviations earlier, improve traceability and make faster operational decisions.
Xpectra combines commercial NIR devices, cloud-based AI models and specialised software. NIR spectroscopy captures a product’s “spectral fingerprint,” while the platform converts that data into visual, traceable quality results that non-expert staff can use.
The system can complement or replace slower, costlier laboratory tests. Because it is non-destructive, the sample remains available, enabling more frequent in-line checks and broader production coverage.
According to the pilot, better classification and management of fish-oil production by quality could increase revenue by up to 12.5% compared with relying only on conventional analysis. This estimate applies to the conditions of the pilot.
Unlike buying an NIR device without tailored models or sending samples to external laboratories, Xpectra offers an end-to-end service: equipment selection, software implementation, calibration, model development and technical support.
AI Talentum provides the technology platform, connectivity and integration; AZTI contributes scientific expertise in NIR spectroscopy, chemometrics, calibration, validation and interpretation of results.
The partners see Xpectra as adaptable to other food products and industrial processes where speed, traceability, analytical reliability and real-time decision-making matter. It supports the wider digitalisation of food quality control, helping companies optimise processes and reduce dependence on external laboratories.
Edited by Jon Erramun Arrue-Karazo