ITMF white paper: data and skills, not algorithms, hold back AI in textiles

Poor data integration, skills shortages and gaps in governance, rather than the technology itself, are the main obstacles to adopting artificial intelligence in the textile and apparel industry. That is the central finding of a white paper published by the Zurich-based International Textile Manufacturers Federation (ITMF) and the Institut für Textiltechnik (ITA) of RWTH Aachen University in Germany.

The paper, titled 'AI in the Textile Value Chain: A Strategic Guide to Artificial Intelligence for the Global Textile Industry', was presented on October 15 at a conference in Fortaleza, Brazil, organised by ITMF together with Brazilian textile and apparel association ABIT and the International Apparel Federation. It builds on a series of ten webinars held between 2024 and 2026, in which more than 20 experts from companies, research institutes and policy bodies took part.

According to the authors, interest in AI is being driven by structural pressures rather than hype, including demographic decline, persistent skills shortages, volatile markets and rising demands for quality, traceability and sustainability. Decades of digitalisation have left the sector with plenty of data but fragmented systems, leading the paper to describe the industry as "not under-digitalised, but digitally fragmented."

From inspection to agentic systems

The paper finds that AI already delivers measurable results in clearly defined applications, such as computer-vision inspection of fabric and yarn, predictive maintenance, energy optimisation and demand forecasting. In one example from a nonwovens production line, causal AI reduced scrap by 10 percent.

At the same time, data quality is described as a bigger constraint than the algorithms. Participants warned that AI amplifies flaws in poor data rather than fixing them, and that many companies hold far less usable, structured data than they assume.

AI is also starting to move from decision support to action. Agentic systems are beginning to coordinate multi-step processes such as sourcing, while causal AI is adjusting production settings directly, in both cases under human supervision. The authors estimate that end-to-end AI workflows are realistic within two to three years.

Trust and collective action

Industry participants said they would only use AI under clear conditions: control over sensitive data, strong cybersecurity, transparency about what systems can and cannot do, and continued human accountability.

The paper recommends that companies invest in people before scaling technology, treat data as shared infrastructure and start with limited use cases. It also sees a strong role for industry associations and pre-competitive collaboration, particularly for smaller companies, through shared data standards, data spaces that leave ownership with the data originator, and sector-specific tools such as a 'Textile GPT'.

"AI is not a shortcut around the structural challenges our industry faces, but it can help companies manage complexity and make the most of scarce expertise," said Olivier Zieschank, director at ITMF.

Thomas Gries, director of ITA at RWTH Aachen University, said companies need a dedicated task force to embed data integration into their technical and business processes, working with external experts on a transparent assessment of their data and automation.