Sustainable Fashion Begins with AI

By: Madhava Venkatesh, Co-Founder and Chief Technology Officer, TrusTrace.

As someone that’s passionate about sustainability, it’s always exciting to see governments step up and do something that matters. Case in point, the European Commission’s Product Environmental Footprint (PEF) program. While still in the testing phase, when it becomes law, it will require brands to calculate and disclose the actual environmental impact of their goods by taking into account supply chain activities: from extraction of raw materials, through production and use and finally waste management. Such legislation would be a windfall for activists that have long pushed big brands to operate more sustainably, none more so than fashion companies.

By widely accepted estimates, the fashion industry accounts for between two and eight percent of the world’s carbon emissions. In 2018, the global apparel and footwear industry alone produced more greenhouse gasses than France, Germany, and the United Kingdom combined.

The PEF is just one of many global regulations that are forcing large companies to account for environmental damage in their supply chains. California’s Transparency in Supply Chains Act and Germany’s recently passed Supply Chain Due Diligence Act are two recent examples. To comply with various new requirements, brands in those regions will need technology solutions for supply chain traceability, plus a new way of thinking about sustainability.

Until recently, brands have had a top-down approach to sustainability, pushing out sweeping corporate initiatives and marketing products accordingly. But this is already a dated and ineffective way of thinking (especially if any real change is going to be made). What’s now required—whether through regulation or an increasingly eco-conscious consumer base —is moving towards sustainability from the product on up.

To produce a truly sustainable garment, brands need to know everything about every product and material they handle. It requires millions of granular, accurate data points and a traceability solution that can house the data in one place.

Why Traceability?

The ability to precisely trace products and materials through the supply chain can help address many challenges. Greater supply chain visibility allows brands to anticipate disruptions before they happen. Plus, such visibility enables brands to make product claims and prove their authenticity. For example, a brand can claim to sell a 100% organic cotton sweater and provide the data to back it up.

As they stand today, fashion supply chains are massive, but with little supplier visibility. Fashion companies therefore face the daunting task of trying to track every product as it moves through hundreds of suppliers across the globe. This reality represents a massive technology challenge that only artificial intelligence (AI) and machine learning can remedy.

AI as a Traceability Enabler

At TrusTrace, we work with dozens of companies in the fashion industry, and much of their supply chain data is locked up in documents—paper and electronic. These documents include invoices that prove the chain of custody, social audit reports describing workplace and pay conditions at factories and other facilities, chemical test reports for material batches and so much more. This document data is often in different formats and languages. In short, the primary issue is data acquisition.

This is where AI becomes critical for traceability. It can intelligently collect mass amounts of data at scale. More importantly, however, it can also  support a system that automatically performs data validation by correlating information from multiple sources to improve the overall quality of traceability data.

More simply, AI can be used to digitize paper trails to enable wholescale product traceability. The digitization process encompasses three steps: Classification, Object extraction and identification, and Data validation and linking.

Classification happens when a document is submitted into a supply chain traceability platform by a supplier. The underlying AI recognizes the document and intelligently classifies it as, for example, a purchase order, facility audit, or certification.

Based on the document’s classification, AI then identifies the key information through metadata. For example, when processing invoices, the traceability system will automatically extract and identify  information like Buyer, Seller, Product, Quantity, Date of Delivery, etc. Similarly, digitizing a social audit might involve capturing parameters related to Working Conditions, Fair Wages, Diversity, and more.

Once the corresponding objects are extracted, the data is validated and linked to other existing data within a brand’s enterprise systems, allowing them to use the data how they wish, whether for forecasting, analytics, regulatory reporting, or other requirements.

Fashion supply chains are so complex and the available data so vast, that it’s virtually impossible to manage without the effective use of AI. After implementing a traceability system, the sustainability of one or more partners in a brand’s supply chain will inevitably fall short of a brand’s standards. In that case, the supply chain must adapt and reconfigure through other partners to remain in compliance. AI and machine learning are the backbone that allows for such rapid adjustment.

Looking Ahead

As the EC’s PEF program demonstrates, there will come a time where it won’t be enough to say you’re sustainable; It won’t even be enough to provide evidence. I believe in a future where brands are calculating in near-real-time how sustainable their products are by intelligently tracing combined materials.

I’m proud to see so many fashion brands committing to sustainability and social responsibility — even before legislation began to ramp up. That corporate commitment must now trickle down to product level. It’s no easy feat, but traceability, backed with AI and data, can make it possible.