Blue Yonder’s Sónia Peres On AI/ML’s Role in Fashion, Retail
An expert in retail tech, Blue Yonder’s Sónia Peres discusses how utilization of advanced technologies such as artificial intelligence (AI) and machine learning (ML) can help fashion firms and retailers adapt to changing consumer shopping patterns, manage their inventories, and get ahead of the curve through trend prediction.
Peres is global vice president retail product management and was previously Blue Yonder’s senior director industry strategies for softline global. Formerly JDA Software Group, Blue Yonder—a subsidiary of Panasonic Corp.—is a supply chain management firm.
More from Sourcing Journal
Sourcing Journal: More and more retailers and fashion brands are now recognizing that AI/ML can be an effective tool in helping them improve operations and drive sales. Why has adoption been slower in fashion retail?
Sonia Peres: AI/ML offers clear advantages in fashion retail, such as enhanced customer personalization, optimized supply chains, and better demand forecasting. However, adoption has been significantly slower in this industry. Traditionally, fashion has been driven by intuition and human creativity, with many resistant to adopting technology, fearing it might diminish the creative process.
Fashion’s subjectivity makes it challenging for AI/ML models to succeed. Individual tastes vary widely, and unlike more structured industries, fashion requires AI systems to interpret not just practical elements like size, color, and material but also abstract factors like style, cultural trends, and personal preferences, which are much harder to quantify. Achieving this level of complexity requires large volumes of clean, structured data. In practice, data on styles, trends, and consumer behavior is often inconsistent or fragmented across sources like online platforms, social media, and in-store interactions. Consolidating high-quality, unified data across these touch points has been a major hurdle for fashion retailers.
Another challenge is the rapidly evolving nature of fashion. Trends shift quickly, and AI/ML systems must constantly adapt in real time. This requires sophisticated algorithms, continuous monitoring, and significant resource investment. Implementing AI technologies also demands considerable financial resources for software, infrastructure, and skilled personnel. For many fashion brands, particularly smaller or traditional ones, allocating these resources poses a challenge, especially with uncertainty around return on investment.
A breakthrough for AI in fashion is its ability to predict emerging trends by crowdsourcing data from social media, identifying trends before they fully form. This capability is now attracting interest from fashion retailers. As the technology matures and industry knowledge grows, adoption is steadily gaining momentum. More retailers are beginning to realize that AI/ML is essential to keeping up with fast-changing market trends and staying relevant to their customers.
SJ: For fashion companies still new to AI/ML, how much, percentage-wise, should technology play a role in the decision-making and how much should be based on other quantifiable data that’s not necessarily a part of the tech algorithms?
SP: Fashion retail will always be a balance between art and science. Initially, fashion companies should allow AI/ML to drive around 60 to 70 percent of decision-making in areas like personalization, inventory management, and trend prediction, where data-driven insights are most impactful. However, 30 to 40 percent should still rely on traditional approaches that involve human creativity, brand strategy, and quantifiable data from non-AI sources. As the company gains confidence in AI/ML, this ratio might shift slightly as technology proves its value in more areas.
SJ: What should fashion retailers spend time on in terms of defining or determining data quality? And what should be the key attributes or pieces of information that fashion companies must include in their data sets?
SP: Fashion retailers must focus on ensuring that their data is cleansed and of a high quality before embarking on any AI/ML initiative.
Key attributes such as detailed product data—e.g., SKU-level details, product attributes, color and fabric data, customer demographics and behavior, sales and inventory performance, and external trend data—should be included to enable effective decision-making. Proper data governance, with regular audits and cross-functional collaboration, will ensure the data quality required for AI/ML success.
SJ: Just when some thought the supply chain disruptions from COVID were a thing of the past, this year saw new disruptions, such as the Red Sea and threats of port closures. What will be the big lesson for fashion retail as they plan for 2025 and beyond?
SP: Supply chains remain vulnerable to unforeseen global events, and in planning for 2025 and beyond, fashion retailers must focus on building resilient, tech-enabled, sustainable supply chains that can withstand both predictable and unforeseen disruptions.
These are some examples of capabilities that should be considered for in building a resilient supply chain:
Advanced technologies like AI and machine learning (ML), real-time tracking, and predictive analytics will play a crucial role in supply chain optimization. Retailers will need to invest in digital solutions that provide visibility into every stage of the supply chain, enabling them to anticipate and respond to disruptions faster.
Building more sustainable, circular supply chains will not only help meet growing consumer demands for eco-consciousness but also help in reducing risks related to resource scarcity or regulatory changes. Agility in production, with the ability to scale up or down quickly, will be critical in responding to both demand fluctuations and disruptions.
Retailers will need to optimize inventory management and explore multi-modal logistics solutions to reduce dependency on any one transportation method. Creating buffer stock for critical items or materials and leveraging decentralized warehousing closer to key markets can help cushion against disruptions.
Strengthening relationships with suppliers and logistic partners will allow for better collaboration in times of crisis. Long-term, collaborative partnerships can create mutual benefits and more flexibility in overcoming supply chain challenges.
Proactive risk management, including scenario planning and stress testing supply chains against various potential disruptions (geopolitical tensions, environmental events, pandemics) will become integral to supply chain strategy. This can help retailers prepare for and quickly pivot during future crisesProactive risk management, including scenario planning and stress testing supply chains against various potential disruptions (geopolitical tensions, environmental events, pandemics) will become integral to supply chain strategy. This can help retailers prepare for and quickly pivot during future crises.
SJ: Since COVID, there’s been greater talk about re-shoring and nearshoring. But the two seem to present bigger hurdles for fashion retailers than companies in other sectors. Any thoughts on how AI can counter this until fashion firms can get factories up and running that are closer to home?
SP: Reshoring and nearshoring have become important strategies for fashion retailers aiming to reduce supply chain risks and shorten lead times. However, these strategies pose significant challenges, such as high costs, lack of infrastructure, and the need to establish new manufacturing capacities.
Until fashion companies can get local or regional factories up and running, [these are the areas where] AI can play a critical role in mitigating these hurdles:
AI-Driven Demand Forecasting and Inventory Optimization
With improved forecasting, retailers can reduce overstocking or understocking, minimizing the risks tied to long supply chains while reshoring or nearshoring efforts ramp up.
Supply Chain Visibility and Predictive Analytics
These AI insights can also help optimize transportation routes and modes, reducing lead times and costs even while production is still distant.
Automated Production Planning and Efficiency Gains
This can be particularly useful for fast fashion, where quick turnaround times are essential. AI can assist with order prioritization and resource allocation, ensuring timely delivery even from overseas suppliers.
Supplier Collaboration and Sourcing Optimization
AI can assist in identifying alternative suppliers or materials in case of disruption, helping fashion companies avoid costly delays while transitioning manufacturing bases closer to home.
Customizing Localized Production Strategies
By analyzing consumer preferences and regional trends, AI can recommend which products to prioritize for localized production, optimizing the impact of nearshoring efforts.
Sustainability and Resource Efficiency
Through machine learning (ML), AI can identify opportunities to reduce waste in the supply chain, optimize fabric utilization, and cut down on carbon footprints even before local factories are established. This helps fashion retailers align with consumer demands for sustainability while they shift manufacturing closer to home.
AI in Product Design and Virtual Sampling
I know that this is still not actively being implemented, but AI-powered design tools and virtual sampling can reduce the need for physical prototypes, which often extend lead times when produced in distant factories, while also eliminating the cost of producing the sample product and sending it out to stores to get a reading on customer preference.
Solve the daily Crossword
