Fraunhofer Institute for Machine Tools and Forming Technology (IWU) has successfully developed an artificial intelligence demand forecasting system tailored for the textile manufacturing sector, marking a significant step toward digitizing production planning. Deployed at frottana Textil GmbH & Co. KG, the manufacturer behind the MÖVE home and terry textile brand in Upper Lusatia, the tool replaces fragmented manual scheduling with a centralized, data-driven platform. The initiative was executed in partnership with Logsol GmbH. The German textile industry, predominantly composed of medium-sized enterprises, routinely navigates pronounced seasonal demand fluctuations. Historically, sales and production scheduling have relied on legacy Excel spreadsheets, isolated employee calculations, and the tacit knowledge of veteran staff. This manual approach has grown increasingly unsustainable due to accelerating skilled labor shortages and age-related workforce departures, which frequently trigger costly planning errors, inefficient capacity allocation, and redundant data transfer between enterprise resource planning systems and offline tools. To address these structural inefficiencies, the IWU engineering team implemented a machine learning architecture centered on neural networks. The system continuously processes historical sales records to automatically detect temporal trends and recurring seasonal patterns. By converting unstructured operational memory into a transparent algorithmic framework, the platform establishes a reliable foundation for monthly procurement and order fulfillment decisions. Crucially, the design maintains a human-centric workflow, enabling planners to validate, adjust, and contextualize algorithmic outputs with industry expertise. This hybrid model ensures rapid onboarding for new personnel and mitigates operational disruption during staff absences. Field testing demonstrates robust predictive performance despite constrained data availability. Using only four years of historical records without granular segmentation by region, sales channel, or promotional activity, the model maintains an average forecasting deviation of approximately nine percent around a baseline of three hundred forty monthly units. The system delivers fully automated, auditable forecasts that substantially reduce scheduling uncertainty while preserving managerial oversight. Following successful validation at frottana Textil, the research consortium plans to embed the forecasting engine directly into end-to-end production planning workflows. This integration will enable dynamic optimization of manufacturing sequences, precise batch sizing, and year-round capacity balancing. The broader initiative aligns with the institute’s SmarMoTEX digitalization program, which simultaneously advances material flow simulation for bottleneck identification, computer vision systems for automated weaving defect detection, and sensor retrofitting strategies to extend the operational lifespan of legacy textile machinery. Together, these interventions aim to modernize traditional manufacturing ecosystems, secure institutional knowledge against demographic shifts, and establish scalable digital infrastructure for the European textile sector.