February 12, 2025
Insight
Predictions & Trends: Future developments in the e-commerce industry assisted by AI - Part I
Written by Ricardo Sousa, PhD (not AI)
SeeOnMe, Chief Artificial Intelligence & Co-Founder
Prediction and Trends
A Q4 2024 report revealed that 74% of companies remain behind in AI adoption, lacking basic capabilities and showing minimal implementation [1] — a concerning trend given AI's growing importance across global markets, right? No. This pattern isn’t new. Q4 2024 report’s findings echo similar adoption patterns seen in previous technological waves, notably the slow uptake of personalization [9] and other AI technologies documented in 2018 reports [8], suggesting a recurring hesitancy in embracing transformative innovations. This is particularly relevant in retail, especially within the fashion industry.
The resurgence of Virtual Try-On (VTON) has been remarkable. Thanks to advances in machine learning architectures (particularly UNet and Transformers), combined with more powerful computing resources (such as the latest TPUv4 from Google, A100/H100/H200 from Nvidia) coupled with almost unlimited access to vast datasets, VTON techniques have achieved unprecedented results.
The technological breakthroughs emerging from these compounded advances across multiple domains offer a glimpse into the future possibilities ahead. What does this new era of AI hold for the fashion industry? Which novel functionalities will emerge, and where will they make the most impact? What possibilities will they unlock, or are we facing another cycle of unfulfilled promises?
Also, what fundamental technological leap - beyond current generative AI - will define the next wave of innovation? In this first article in our series, we will offer a concise yet thoughtful insights into AI's role in fashion, examining its future implications.
AI Disrupting Retail World
Virtual try-on technology has evolved from its basic origins to a pivotal turning point. While initially focused on basic visualization, today's VTON solutions are approaching photo-realistic rendering and accurate fit prediction which will soon be widely available to the end user. With e-commerce representing 20.5% of global fashion sales, this advancement promises to solve retail's persistent returns challenge — potentially revolutionizing how we shop for clothes online. Solving this technical challenge could finally address the customer's fundamental need for personalized fit and wear.
Personalization
As AI is enabling enhanced visualization of materials and patterns based on consumer preferences, it will transform traditional personalization into hyper-personalized experiences [6]. In practice, hyper-personalization extends traditional data analysis by combining diverse data streams across platforms and providers — from social media interactions to shopping behaviors and lifestyle preferences — while leveraging multi-modal inputs including text, images, and videos. This rich combination of data sources, processed through advanced AI capabilities and robust cloud computing infrastructure, enables deeply personalized experiences for each customer.
Marketing
Following suit, one will see an extensive use of AI tools to empower marketing teams by creating more targeted and cost-effective campaigns through advanced data analysis. By enabling the rapid generation of diverse content across photos and videos, brands can experiment with multiple creative approaches whose value will be ascertained through A/B testing. Building on existing successes [5], such will allow teams to understand user preferences, leading to more engaging and efficient marketing strategies [4]. As marketing campaigns become increasingly integrated with preference-based strategies, deeper insights into customer behaviors emerge, opening new possibilities for creative development.
Production
As a matter of fact, we are already witnessing a paradigm shift in clothing design. Digital sketch-to-photo conversion, while established in fashion design, is gaining new applications through technological advances that enhance the creative process. The multi-stage creative process will feature AI-assisted iterative design, enabling continuous refinement. This transformation extends beyond creativity to revolutionize e-retail product catalogs, accelerating the catalog generation process through instant generation of apparel displayed on diverse virtual models, dramatically reducing time spent on photo shoots and editing. Moreover, these tools democratize fashion design, enabling individuals with minimal technical expertise to become creators, effectively lowering the barriers to entry in digital fashion design and merchandising.
Simulation will mark a pivotal shift in the textile industry, empowering creators and manufacturers with advanced predictive capabilities. Through platforms that generate multiple scenarios of fabric behavior—including folding patterns, wrinkle formation, and size grading—designers can accurately forecast real-world garment performance before production [12]. This technological advancement not only streamlines the design process but also enables marketers to create compelling, true-to-life promotional content, bridging the gap between digital visualization and physical product reality.
Bricks and Mortar
Our team has been dedicated to augmenting experiences in the fashion world for over a decade and has been studying and following its various transformations closely. While COVID brought a sense that physical purchases were becoming obsolete, we have seen a steady return to physical stores in the post-pandemic period. We won't delve into the reasons behind this trend, but instead focus on whether VTON will strengthen the thesis that online presence will prevail. While VTON helps bridge the gap between the real and online world, there are three key considerations to keep in mind. First, haptics - the capability to remotely sense something from the physical world in the virtual realm - remains distant; secondly, engagement and human expert guidance continue to prevail during considerable purchases, specifically luxury goods; third, foundational disruptive technologies such magic mirrors that continue to transform in-store experiences. Through the convergence of tactile feedback systems, personalized virtual experiences, and enhanced in-store visualization technologies, retailers can forge a more seamless integration between digital and traditional shopping environments.
Copyright
As the fashion industry evolves, established brands must prioritize intellectual property protection and brand integrity in the digital realm. The emergence of sophisticated AI detection systems, alongside embedded digital watermarking technologies, will become essential tools in combating counterfeit designs and unauthorized brand appropriation. This technological safeguarding, coupled with regulatory frameworks similar to plagiarism detection systems, will help maintain brand authenticity and protect creative assets in an increasingly AI-driven fashion ecosystem.
These technological advances naturally promote fashion sustainability by reducing waste and optimizing resources across the entire value chain [4]. To which, all of them can support product development by analyzing large social media and runway show datasets to identify emerging fashion trends.
And, consumers will follow suit as they embrace the benefits technology offers. While these disruptions may reshape traditional retail [7], businesses must adapt to this revolution or risk becoming obsolete. Companies that harness their value chain benefit from both reduced operational costs and improved conversion rates through streamlined systems.
Final Remarks
First, a note of both caution and opportunity is warranted: While most companies lag in AI adoption [1], a select group beyond the 'magnificent 7' [2] is rapidly advancing, potentially creating both opportunities and unexpected challenges other contestants. Some of those encompass technical debt and scalability which could lead to operational inefficiencies [3].
This conundrum of early adoption versus technical debt presents a unique opportunity for emerging startups, who can not only build their business models with these lessons in mind but also leverage their advantage of starting from a clean slate. Without legacy systems to maintain or outdated architectures to accommodate, these new ventures can implement cutting-edge solutions from the ground up.
An exceptional opportunity for partnerships between emerging startups and early adopters. Organizations can actively shape tailored solutions that address their immediate needs while testing innovative approaches without the burden of legacy constraints [7]. Retailers looking to enhance their online customer experience with SeeOnMe's API can connect with our team directly at team@seeonme.ai.
Our next article will explore the trends and predictions on emerging technologies driving fashion's transformation and their potential future developments.
References
Sources used to support the content in this article.
https://www.investors.com/research/magnificent-seven-stocks-january-2025/
https://the-cfo.io/2025/01/17/the-roi-puzzle-of-ai-investments-in-2025/
https://cx-journey.com/2024/07/moving-beyond-personalization-to-hyper-personalization.html
Notes from the AI frontier: AI adoption advances, but foundational barriers remain in 2018 McKinsey&Company
https://www.deloitte.com/uk/en/services/consulting/blogs/2020/failure-of-personalisation.html
https://www.forbes.com/sites/zengernews/2024/07/11/virtual-try-ons-will-change-fashion-jobs-forever/
https://www.weforum.org/stories/2016/06/industry-4-0-business-in-the-age-of-personalisation/
Li, Yifei, et al. "Diffcloth: Differentiable cloth simulation with dry frictional contact." ACM Transactions on Graphics (TOG) 42.1 (2022): 1-20.
For future reference
