Sajna Massey, founder of Toronto-based Couth Studios, launched the brand in 2024 to tackle fashion overproduction using artificial intelligence and data-driven design.Nick Kozak
The fashion industry has a waste problem.
Forty-six billion garments are manufactured each year but never worn, according to Oxfam research. That’s 40 per cent of global clothing production going straight to landfills.
Sajna Massey, a software engineer with no fashion industry experience, believes she has a solution.
Ms. Massey spent years at an artificial intelligence venture studio working across medtech, legal tech and education tech, helping machine learning transform diverse sectors. Fashion, she thought, could benefit from the same tools. She saw an opportunity to combine her love of clothes with her passion for sustainability. In March 2024, she launched Couth Studios, a Toronto-based brand designed to tackle overproduction in the clothing industry with AI.
The brand uses AI to generate designs, but the technology is only part of the process. Rather than simply automating creative work, Ms. Massey built a system that combines AI-generated designs, data scraping and community voting through Instagram polls. The approach aims to ensure clothes get made only if there’s demonstrated demand.
The process starts with her custom-built web scraper, which pulls data from runway shows around the world. The system analyzes photos to identify specific physical attributes and calculates how frequently these elements appear across collections.
“As an example, it’ll say: 7 per cent of clothes have oversized shoulder pads, 8 per cent use gold hardware or 10 per cent have an asymmetrical silhouette,” she says. That data feeds into a design algorithm that generates new pieces, primarily using Midjourney, the AI image generation platform.
But the AI-generated designs don’t go straight into production. Ms. Massey first posts everything on Couth Studios’ Instagram page, where her community votes on which pieces they want to see made.
Massey uses AI tools, Instagram polls and small-batch production to ensure clothing is only made when there’s proven demand.Nick Kozak
It’s a departure from traditional fashion’s top-down approach, where designers present finished collections and hope consumers respond. Ms. Massey inverts that model, letting demand shape production from the start. After running a series of 10-week voting cycles, she produces only the top performers in small batches.
She says the AI-generated designs require significant human intervention before she presents them to her community. She employs a part-time team member to validate the designs and works with local designers on pattern-making, sample-making and refining constructions for efficient production. The review matters, she says, because AI can generate impractical designs.
“A dress might cost $1,000 to produce or it’s something else that isn’t really going to sell,” she says. “It’s pretty rare that the AI generates something and we keep it exactly as it is.” Even in an algorithm-driven process, there’s always human originality in the design.
That human curation addresses some of the ethical questions Susan Langdon, chief executive officer of Toronto Fashion Incubator, says designers face as AI becomes more widely adopted in the industry.
“Are you being authentic to your audience if a computer generated your design?” she asks. “Are you infringing on someone’s copyright or likeness without their permission?”
For Ms. Massey, the answer lies in maintaining creative control throughout the process. She focuses on wearable designs that feature statement silhouettes with solid fabrics, with every AI-generated suggestion vetted against garment guidelines.
“We don’t want to be in the business of selling clothes that people only wear once,” she says.
Ms. Langdon says Canadian designers are experimenting with generative AI for lookbooks and hero images, using platforms such as Runway and Midjourney. But she believes adoption will soon accelerate even more out of competitive necessity.
“Canadians will be forced to catch up with the rest of the global fashion industry in order to be competitive and embrace AI in all aspects of their business, from concept to sampling, marketing and production,” she says.
She points to services already available in East Asia where designers can have AI create virtual samples for e-commerce sites, develop social media strategies and send orders directly to contractors who drop-ship to customers.
“This approach is extremely cost- and time-efficient, as the sampling process is one of the most expensive parts of research and development,” she says. “It’s not available in North America yet, but it’s coming.”
As these systems become more accessible, Ms. Langdon adds, “The challenge for Canadian designers will be finding the right balance between efficiency and innovation, and authenticity and technology.”
Then there’s the environmental paradox. AI has drawn criticism for its energy consumption, yet Ms. Massey is using it to solve a waste problem. AI has its own footprint, but generating digital mock-ups can be comparatively light, with one study estimating a cost of roughly 2.9 kilowatt-hours per 1,000 AI images. That’s less than 0.003 kWh per image. For context, charging a phone once daily uses approximately 0.035 kWh of electricity per week. Ms. Massey has weighed the trade-off and believes it’s worth it.
“The problem we’re trying to solve is bigger than the energy we’re using to solve it,” she says.
There’s also one area where technology may not yet be able to replace human touch. Despite the growing popularity of virtual sampling around the world, Ms. Massey says it “just didn’t work” when she initially attempted that approach. “It’s tough to convert that way,” she says. “There’s so much noise online.”
Now she makes the top six voted designs into physical samples first, then hosts local events where people can see, feel and try on the clothes before pre-ordering.
“I think we needed that in-person component,” she says.
Whether AI-driven, community-voted fashion represents the industry’s future or a niche experiment remains unclear. But Ms. Massey’s experience so far suggests many Canadians still want to examine the clothes up close before they’ll buy. The technology may be cutting-edge, but the final sale remains decidedly analog.