THE IDEA BEHIND THE WORKSHOP
This project is a creative, hands-on AI design workshop developed for fashion designers, SMEs, educators, and students. It aims to help participants explore and apply AI tools in a practical and inspiring way—boosting confidence, creativity, and innovation. The workshop is supported by an insight video, a website with detailed process, and optional consulting to guide brands through early adoption.
To move beyond a theoretical understanding of the role of artificial intelligence (AI) in the fashion design process, it is essential to explore its practical application within the industry. Particular attention is given to small and medium-sized enterprises (SMEs), where limited budgets and resources often pose challenges to adopting new technologies, despite their growing relevance.
As a Fashion & Business student, I believe the early stage of product development — the design phase — is one of the most important areas in a fashion company. It is a space driven by creativity, intuition, and vision, much of which cannot be taught or fully automated. However, with the growing presence of AI, this creative process is beginning to shift. Designers are now being asked to rethink their role — not as individuals working in isolation, but as collaborators with intelligent tools that can assist, accelerate, and even inspire new design outcomes.
This workshop was designed to explore the real-world feasibility of using generative AI tools to create fashion graphics that align with a brand’s seasonal direction and visual identity. Drawing from my internship experience at Stitchd, the workshop focused on whether AI can truly enhance the design process or whether it remains a trend-driven idea lacking practical value. Participants were invited to engage directly with AI tools, testing their ability to contribute to concept development, overcome creative blocks, and support visual experimentation.
Through this hands-on approach, the workshop aimed to provide insight into both the benefits and limitations of AI in a fast-paced design workflow. It also served as a platform to reflect on broader industry questions: What does creativity look like in the age of AI? How much of the designer’s role can or should be supported by digital tools? And what new skills are needed for designers to thrive in this changing landscape?
Ultimately, the goal was not only to test AI's technical potential but also to encourage critical thinking and creative exploration, helping designers and teams better understand how these tools fit into their evolving roles — and whether they can truly add value to the future of fashion design.
The AI Experiment with the design team at Stichd
This section presents a hands-on experiment conducted in collaboration with the design team at Stichd.
Stichd is a product licensing company and subsidiary of the PUMA Group. Stichd partners with some of the world's most iconic brands, including Levi's, Calvin Klein, BMW, Formula 1, and Manchester City FC. The company is primarily known for designing and manufacturing licensed products across various categories, including fashion essentials (Legwear, Bodywear, Swimwear), Fanwear (Motorsport and Teamsport), and Lifestyle apparel and accessories.
The experiment followed a five-stage design thinking framework to explore how generative AI tools can support the design process. Designers were invited to use selected AI platforms to generate print patterns and concept visuals. The process involved individual exploration, team sharing, and group discussion to assess both the experience and the outcomes.
The experiment highlighted both the creative benefits and the challenges of integrating AI. These findings are crucial in assessing the feasibility and practical value of AI-assisted tools in fashion design.
Step-by-step Process of the Experiment
DESIGN INSPIRATION from LEVI’S brand
YOUNIFORM: Inspired by the new season and the energy of returning to the city, YOUNIFORM reimagines normcore through modern prep classics. The collection delivers effortless, versatile pieces grounded in individual style—romantic layers for women and subtle sportswear nods for men.
Classic Clash: A story driven by style tension.
In menswear, classic prep is reworked with rugged outdoor and military influences, grounded in a refined yet rebellious spirit—punctuated by flashes of Ivy Sleaze.
In womenswear, reinterpreted prep collides with romantic revival, layered and dimensional, shaped by proportion play and echoes of the '80s and 2000s.
New Romantics: Unfiltered style meets high-low contrast.
New Romantics merges gritty and glam, channeling the spirit of an indie renaissance.
This aesthetic is brought to life through cozy, dimensional textures and soft layers, expressing both edge and emotion with modern romanticism.
Step 2: Set Design Context
The design goal was to create new graphic concepts that aligned with the brand’s identity, reflected the seasonal direction, and remained commercially viable. A key challenge was to develop fresh, original, and market-ready graphics while staying true to the brand’s signature style and within the defined price limitations.
The team was encouraged to push beyond existing design references and generate ideas that felt newer and more innovative than in previous collections—an ambitious objective that made the task even more demanding.
** Image blurred intentionally to protect confidential brand materials.**
STAGE 1 EMPATHIZE: UNDERSTAND THE GOAL OF IMPLEMENTING AI
Step 1: Planning & Objective
To align with upcoming seasonal collections for a brand that collaborates with Stichd, we began by receiving the initial design direction and inspiration from a brand. The Stichd design and product management teams, along with product development, held a briefing session to discuss the creative direction, quality expectations, and price points that the designs should meet. This also included an analysis of past sales performance to inform what kinds of designs tend to perform well and worst to avoid in the next collection.
Step 3: Competitor, Market, and Trend Research
As part of the research phase, the design team examined competitor collections and conducted market trend analysis, with a focus on key global markets. The objective was to gather insights and inspiration from emerging styles, consumer behavior, and broader market patterns. While an AI trend forecasting tool could have added value by filtering and visualizing potential directions, it was not utilized due to access limitations and associated costs. As a result, only traditional observation and manual research methods were employed during this stage.
** Image blurred intentionally to protect confidential brand materials.**
STAGE 2 DEFINE: CLARIFYING DESIGN INTENT
Step 4: Introduction to Generative AI Tools
After the design direction was set and market research insights gathered, generative AI tools were introduced as part of the creative ideation process. While the tools were new to most of the team, the researcher began experimenting with platforms such as Midjourney, DALL·E, and Runway to explore their potential. The team continued to rely primarily on traditional methods, but AI was tested as a complementary support for idea generation.
Colour layer 3
Step 5: Prompt Crafting
A key learning curve involved understanding how to craft effective prompts. The researcher explored how specific, detailed language, particularly the use of technical terms and design knowledge related to styles, materials, and patterns—led to more accurate and relevant results. The findings highlighted that well-constructed prompts play a critical role in generating higher-quality, more efficient outcomes from AI tools.
"AI speaks in data — designers speak in vision." - Insight from researcher
To develop a fresh and fashion-forward pattern direction aligned with the Stitchd goals, AI generative tools from different platforms were used to explore visual concepts based on the following creative prompt:
Prompt 1: The design features a bold, motion-blurred rectangle in cream white, with subtle glitch and topographic textures in the background. Around the blurred form, add delicate stitched contour lines in tonal thread, mimicking topographic maps. Include extra stitched elements such as dashed motion streaks, offset grid lines, and small embroidered symbols like a compass, tree icon, or signal waves. The pattern includes layered organic elements like blurred leaves, rock textures, and shadowy forms in earthy tones—olive, sand, khaki, and muted black. The design is expressive, painterly, and experimental, evoking a contemporary techwear or urban camouflage aesthetic.
Prompt 2: Seamless repeating camouflage pattern blending abstract photographic textures with digital distortion. The pattern includes layered organic elements like blurred leaves, rock textures, and shadowy forms in earthy tones—olive, sand, khaki, and muted black. The design is expressive, painterly, and experimental, evoking a contemporary techwear or urban camouflage aesthetic.
The picture shows AI-generated output based on custom prompts created by the researcher by Midjourney platform
STAGE 3 IDEATION: GENERATING IDEAS
Step 6: Tool Selection & Exploration
In this step, I explored and compared three AI tools — Runway, DALL·E, and Midjourney — by using the same prompt across each platform. The aim was to understand how each tool interprets creative direction differently and to assess their strengths, limitations, and overall suitability for fashion design, particularly in concept generation.
Each platform generated a distinct result, offering insight into how tool selection directly influences visual style, usability, and workflow integration. These differences are especially important for fashion designers experimenting with AI, where both visual aesthetics and technical readiness matter.
Runway delivered the strongest overall result in terms of style and design quality. The visuals were dynamic, balanced, and closer to production-ready outcomes. However, Runway’s premium pricing model makes experimentation more limited. To make the most of this tool, designers need to have a clear concept and desired outcome in mind, as trial-and-error can become costly.
DALL·E ranked second. It offers unlimited free experimentation, making it a great option for early-stage ideation. The outputs showed clear object recognition and clean visuals, though they leaned toward simpler, less stylized results. It’s easy to use, making it beginner-friendly for those new to AI image generation.
Midjourney produced the most visually rich and artistic outputs, with strong textures, abstract compositions, and experimental aesthetics. However, its results were often not technically practical for fashion design production — images lacked proper formatting, layering, and resolution. It serves best as a creative inspiration tool, helping to explore mood, tone, and visual direction.
This comparison highlights the importance of selecting the right AI tool based on the specific stage of the design process and the creative objective.
Runway is ideal for final concept refinement due to its high-resolution output and polished visuals, making it suitable for near-production-level development.
DALL·E is best suited for rapid experimentation, offering unlimited prompt testing with clear object recognition and accessible results.
Meanwhile, Midjourney excels at generating inspirational visuals, making it a powerful tool for moodboarding and early-stage creative exploration, even though its outputs may not be technically production-ready.
Once the researcher gained a foundational understanding of how each tool worked, the process of generating visuals began. The focus was on print and pattern development, mood exploration, and graphic experimentation for bodywear and legwear categories. Many designs were generated but not all of them answer to the right design direction.
"The real act isn’t clicking ‘generate’—it’s knowing what’s worth keeping." - Insight from researcher
The three images show how the same prompt can generate different results, with each tool producing unique styles and quality.
STAGE 4 PROTOTYPE: REFINING AND EVALUATING DESIGNS
Step 7: Sharing & Team Feedback
During a design briefing meeting, the researcher presented a selection of prompts, AI tools, and outcomes from the generative process. The discussion included both technical and creative challenges encountered along the way. The Stichd design team reflected on the originality of the outputs, their feasibility for production, and how well the AI-generated visuals aligned with the brand’s identity and design direction.
Several results were positively received. Out of 60 AI-generated designs, 2 were selected for further development as part of the upcoming collection options. Other outputs served as visual inspiration, helping to guide additional design exploration.
"AI might generate thousands of options—but only the designer knows which one feels right."
- Insight from researcher
This design has potential for the upcoming collection; however, the fine line details are not suitable for the printing technique, as they would result in poor quality. While inspiring, the design is not technically feasible. This highlights a key limitation of AI tools—their lack of technical awareness when it comes to production processes and material constraints. To move forward in the print production process, this design was further refined by the designer.
Colour layer 1
Colour layer 2
This design could not be completed solely by the generative AI tool. While the AI generated a multi-colour print, producing it for manufacturing requires the colours to be separated into distinct layers—something the AI was unable to provide. As shown in the colour layer images, the generated layers were not correctly structured. The details had to be manually refined and finalized by the designers to ensure production feasibility.
STAGE 5 TEST: ITERATING FOR PRODUCTION FEASIBILITY
Step 8: Refinement & Curation
Designers combined the selected AI-generated visuals with traditional sketching methods and manual techniques. This phase demonstrated that AI can act as a valuable starting point in the creative process, rather than serving as a complete design solution.
"Presenting the AI workflow helped bridge the gap between creative concepts and production possibilities."
The two designs below demonstrate how AI still lacks the technical understanding required for production-ready outputs.
Almost 60 visuals were generated by AI tools in just five hours, with results that were both satisfying and creatively inspiring.
TEAM FEEDBACK & KEY TAKEAWAYS
What Worked Well:
Sparked new inspiration and creative ideas
Some designs felt fresh and commercially strong
Enabled customization of different design directions in a short time
Supported alignment with brand identity
Accelerated visual generation—hours instead of weeks
Colour layer 4
What Didn’t Work:
Limited understanding of how the tools functioned
Final designs still required manual refinement in Photoshop
Many generated outputs were not usable
Low resolution and incompatible file types (e.g., no PSD files for production)
Some designs were not technically feasible for production (e.g., too detailed for print)
Raised concerns around authorship and copyright
What Surprised the Team:
Prompting is a critical skill—designers need to know technical terms
The tool subscriptions are affordable, but using multiple tools adds up
Impressed by how quickly good-quality results could be generated
Afraid AI might replace or reduce human jobs
Expert voices from Stichd
“I need more creative stimulation beyond just trend and market research. Experimenting with generative AI tools during the assignment gave me fresh visual inspiration, which I could then develop into new graphics and design directions.”
“We can’t ignore this tool—it has the potential to give us a real advantage across many aspects of the design process.”
“It’s not only about creating the design—it’s also about how we present it. With AI, we can visualize the product more clearly for the market or the head team, helping them understand the concept instantly.”
“I know many brands have already integrated AI into their creative and product development processes. For Stichd, it’s still something new—as most of the work is currently done manually. It would be great if we could begin exploring this too—not just to follow a trend, but to show how these tools can bring real value to our design team.”
“AI tools can be beneficial, but you need to really understand how they work.”
“I need to start learning about the tools. I don’t want to be left behind or feel like they’re going to replace my job in the future.”
Personal Reflection & Conclusion
"Rethinking Fashion Design in the Age of AI"
As the researcher and intern at Stichd, this workshop allowed me to explore the practical potential of generative AI in the fashion graphic design process. I discovered that AI tools can significantly accelerate early ideation, helping designers visualise new concepts quickly and from unexpected perspectives. In just a few hours, nearly 60 unique visuals were produced, an output that would typically take days or even weeks using traditional methods. This speed is especially valuable in today’s fast-paced fashion industry, where quick decision-making and concept development are essential.
However, the experiment also revealed important limitations. Many of the AI-generated visuals lacked technical readiness for production. Files were often low-resolution, lacked proper layers, and did not account for material or printing constraints. These shortcomings required human designers to step in and refine the outputs manually. Additionally, some team members expressed uncertainty about how to use the tools effectively and raised valid concerns about creative authorship and originality.
Through this experience, it became clear that AI should not be seen as a replacement for designers, but rather as a creative partner. The most successful outcomes emerged when human input shaped, adapted, and aligned AI-generated visuals with brand identity and production needs. Another key learning was that the choice of AI tool is just as important as the creative prompt itself. Understanding the strengths and limitations of each platform helps designers make more informed decisions, ensuring that the tool they choose aligns with their creative goals and technical needs. The wrong tool can lead to weak results or extra work, while the right one can streamline the process and improve outcomes.
For AI to be truly useful in a design setting, teams need time to explore the tools, understand how they work, and learn how to craft effective prompts. Successful adoption depends not only on access to technology, but also on having the right mindset, training, and space for experimentation.
For small and medium-sized fashion companies, I believe AI can offer real value, especially during the early stages of the design process. Yet many designers in these businesses remain hesitant. Some view AI as too technical or unrelated to their workflow, while others feel unmotivated to explore it.
This hesitation highlights the importance of creating accessible entry points, such as workshops, toolkits, or digital guides that allow designers to experiment with AI in a practical, low-pressure way.
From this perspective, the value of my research goes beyond academic insight. I present this experience as a workshop format that empowers designers and SMEs to explore AI with confidence and curiosity. This takes the form of a guided, hands-on session and a digital presentation that combines design thinking with practical understanding of AI tools.
In conclusion, this workshop confirmed that AI has the potential to be a meaningful addition to the fashion design process. When used with the right mindset and skills, it can support creativity, improve efficiency, and open new directions for design — while still relying on the human touch that gives fashion its soul.