ATMPs require specific controls and procedures when they are produced, unlike classic drug manufacturing. Characteristics like the inherent complexity around transforming biomaterial into personalized drugs, the variability of each batch highly dependent of multiple external factors and the need of not having margin for errors, make this new way to produce medicinal products an excellent scenario for AI. Techniques like dimension reduction, pattern recognition, anomaly detection and clustering are usually applied in biopharma manufacturing nowadays with outstanding results. The same principles are applicable to ATMP establishing the right redimension from a non-linear perspective. These approaches will be described during the presentation including use cases that would help to understand the benefits of this AI application.