What problem originally drew you into building Mercury Dasha?
Fashion product development has always been really inefficient and this causes a lot of wasted costs and wasted materials. I personally experienced this because I had a small London-produced womenswear brand. And I know first-hand the expense, waste and lack of ability to reduce risk for small founders and market entrants largely because we really struggle to know what to produce and because there is so much upfront cost. This can all now be massively transformed and optimised using technology. We have traditionally relied too much on best guesses and this was as a result of a number of factors. Collaboration has been too fragmented across key stakeholders, lead times are really long and risk lagging beyond current demand and getting commercial feedback far too late means that product decisions are still made on instinct and with imperfect information. What drew me in was seeing how much creativity, time, and capital was lost because of these fundamental issues with the way products are conceived, developed, produced and marketed in the industry. I wanted to build something that respected creativity while doing my part to see if i could help creators to succeed in a difficult industry.
You talk a lot about being a product-led founder. What does that mean in practice?
It means starting with a real, persistent problem and letting the product evolve around solving it — rather than chasing technology trends or investor narratives. In practice, that meant avoiding the AI hype cycles and instead seeing current technologies as a tool to improve how fashion products are actually created and brought to market. It means rather than thinking about how you will reach customers, instead focusing on how they will interact with your product and letting the product lead the way to growth. For me, being product led is very closely tied to my values which is that I place a high priority on innovation and user experience above all else.
Why did you choose to build Mercury Dasha in phases instead of trying to solve everything at once?
Because in order to build a valuable, multi-dimensional product you need to grow through phases of learning and that can only come when you test and then build more and more of what unlocks value. You may start with something superficial but as time goes on you find that to unlock true differentiation you need to drill much deeper into where the value is for the user. Early versions helped us understand where intelligence actually mattered and where it didn’t. V1 and V2 taught us that design generation was useful, but insufficient on its own. Each phase unlocked deeper insight into the real problem. Building in phases wasn’t about moving slowly; it was about building depth and defensibility
Many AI tools in fashion focus on generation. Where do you think that approach falls short?
Generation solves only one moment in a much longer lifecycle. Fashion fails when designs ignore cost, production reality, or demand signals. Most AI tools stop at attractive outputs. They don’t consider how it fits into the wider system. What we are doing isn’t about producing more design options — it’s about helping teams choose better ones, within a certain context. Also fashion is very different to art in the sense it must be produced to be used - so if you want to create more than fashion-art or digital fashion, you need to go beyond generative image AI for design exploration.
You’ve been careful to built Mercury Dasha in a way that distinguishes it from other superficial image generation tools. Why is that distinction important?
Because many AI companies are built on prompts alone - and only really offer an interface or 'wrapper'. This is not defensible. If the value of your product lives entirely in a prompt, anyone can replicate it. What makes Mercury Dasha different is that the outputs are shaped by interconnected inputs we own and structure — brand DNA, feedback signals, constraints, and collaboration data. Even if we continue to use foundational models, and even if we use prompts, the intelligence lives in the system or workflow, not the prompt. At that point, the prompt becomes a function, not the product itself.
How does intelligence actually improve over time inside the product?
Every interaction creates signal. Decisions made, feedback given, constraints applied — these aren’t just events, they’re data. Over time, the system learns what works for a specific brand, team, or context. That cumulative learning is what makes the product harder to replace. The longer you use it, the more tailored and relevant it becomes.
Fashion is subjective by nature. Are you trying to predict what will sell?
No — and that’s intentional. Fashion should and can never be 100% predictable. Surprise and intuition are part of what makes it compelling. Our goal isn’t certainty; it’s confidence. We help teams make better-informed guesses by combining creative intent with signals and constraints. Intelligence should guide decisions, not eliminate creativity.
What has been the hardest part of staying focused as the product evolved?
Accepting that clarity comes through building, not before it. There are moments in product development where everything feels messy and uncertain. What kept me grounded was a clear mission: helping creators succeed without compromising their creative point of view. That mission became the filter for every decision and helps me stay focused when things become confusing.
What lesson would you share with founders building AI-first products today?
Don’t confuse technology with value. AI makes execution faster, but it doesn’t replace what has always been the core question of entrepreneurship - are you solving a hard problem and would people actually thank you for what you have created? The real work is not execution anymore - implementing AI is available for everyone now - it is actually applying domain experience and deep insight into the user experience and then thinking about how AI assists with that - in other words, as always, leading with the problem not the technological solution.
What has building Mercury Dasha taught you about entrepreneurship that you didn’t expect?
I think I underestimated the importance of having a clear mission - because your product may go through multiple pivots and iterations but as long as you have the same mission you will feel grounded in that. What I’ve learned is that strong founders hold the direction steady while allowing the shape to change. Building the product revealed truths no deck or roadmap could. Entrepreneurship is less about certainty and more about learning faster without losing your values. And that is why its really important to have a really firm sense of mission - and I have always had that. It must be really hard to stay the course, navigate decisions and hardship without that anchor.
What do you think is the biggest challenge facing fashion students and new graduates today?
They’re entering an industry with higher expectations, faster than ever and fewer safety nets. They’re expected to be creative, commercial, technical, and digitally fluent — often without access to the systems that make that possible. Many graduates have talent but lack context: how designs translate into revenue. That gap between education and reality can be where many promising careers stall. Many schools are still not teaching enough technology, although many are open to it, they still have a resistance to adopting new methods because they thing it dilutes the craft and I think ultimately this puts students at a market disadvantage. Its important to have a base in both traditional and emerging methods.
Do you think AI is making it harder or easier for young creatives to succeed?
AI lowers the barrier to experimentation and to entry, which is empowering. But it also raises the bar for originality and judgment. When everyone can generate outputs, the differentiator becomes taste, decision-making, and point of view. That is something AI and data can help with but cannot fully replace. Also if you are starting a brand the design part is just one element, you must understand how to market and sell your products, brand yourself and communicate your differentiating value - and get products made efficiently. We help with all of these elements and make it easier, cheaper, quicker and more efficient for young creatives to break through. However ultimately you still need talent and hard work as its a competitive industry.
How do you balance creativity and commercial reality without compromising either?
Creativity should express an opinion. Especially in fashion there must be a cohesive point of view - following AI blindly simply based, for example, on what may sell, will leave you rudderless. Commercial intelligence should inform feasibility and risk, not dictate taste. When those two are connected properly, teams gain freedom — because they understand the trade-offs earlier and can choose deliberately. AI, in agentic forms, can suggest, and even, over-time, create autonomously, but the curation and selection role must be done by a human with a creative viewpoint.
What do you think most early-stage founders misunderstand about product defensibility?
They assume defensibility comes from features or technology alone. They also think its because of some super secret IP or patent. In reality very few companies develop that sort of formalised IP.. In reality, defensibility comes from accumulation of data, learning, workflows, and trust built over time. Defensibility is a byproduct of sustained problem-solving. If your product gets meaningfully better the more it’s used, and is used by more and more people, you’re on the right path to building a defensible product.
Do you see Mercury Dasha’s approach applying beyond fashion?
Yes. Fashion is a complex creative industry with tight margins, long lead times, and subjective demand. If you can build intelligent workflows there, the underlying approach applies to adjacent sectors like homeware, furniture, consumer goods, or even industrial design. The common thread isn’t fashion — it’s products that sit between creativity and manufacturing that have complex and long product development cycles.
What would you say to creative students preparing for an AI driven world?
I think its still important to teach craft because even if AI can do it quicker - you still want to be able to critique, improve and sometimes even overwrite it. AI has a style of image generation that depends on the model - so your unique handwriting is still essential and you should take time to develop that as this is what AI will interpret. If you don't do that then you are at risk of producing outputs that look generic. So as well as the craft you should also immerse yourself in inspiration that creates your original, creative viewpoint. Education should also teach context. Understanding how various business units work together and how creative decisions impact on commercial and vice versa is just as important as aesthetics. And also sales skills, because even if you are a creative you still need to market yourself and your work.
If you were starting again today, what would you do differently?
I am not sure I could have done anything differently because every entrepreneur, I assume, does the best with the information and resource they have. One thing I would do is appreciate how long everything takes including raising capital and adjust accordingly. Your runway is always much shorter than you'd like it to be!