How Retail Brands in India Are Using AI to Grow Faster | Farooq Adam, Fynd

An Indian brand takes 150 days to get an idea onto a shelf

Farooq Adam, co-founder of Fynd, on why the fashion calendar broke, what happens when you hand a designer a hundred options instead of ten, and the 1,200 factories sitting behind it.

If a film becomes a trend on Monday, a brand working on a conventional supply chain cannot have a collection responding to it. Farooq puts the Indian cycle at 150 to 200 days from idea to market. Global sourcing chains can run to a year and a half, because they are trying to aggregate demand worldwide and predict two years ahead.

Nobody sitting in 2020 predicted 2020. That is the problem with a forecast that far out, and it is what Fynd Create was built to compress. Farooq spoke to Ragini about design explosions, what small-batch manufacturing did to factories that didn’t ask for it, and why he thinks the creative process itself doesn’t change.

"Customers are moving faster than our calendars."

The work around the work

A high-throughput designer might produce a few dozen designs a month, and that isn’t just the design, it’s the tech pack end to end that a manufacturer needs. The process is largely manual. Pen and paper, then illustration, then tech pack software, then email or WhatsApp with your vendors.

So you have data sitting in multiple places and no view of how any of it comes together. I call it software drudgery. The real work of a designer might be a few hours, and then the work around the work is bigger than the outcome itself.

A design explosion, then a human filter

In 2022 the AI models became good enough to generate genuinely new designs. The idea was simple: an idea in your head is a prompt. What if that becomes real output the brand selects from, rather than the designer drawing every option? Instead of ten designs you make hundreds.

Our own designers’ productivity has gone through the roof, and their value hasn’t changed at all. They still decide what gets made and how. We are making products for humans, and humans have to decide what humans like. No matter how much data or how many AI systems you have, someone still decides. AI can produce thousands almost instantly. It’s the curated ones that go to the brand.

Learning a brand means learning what not to do

Most of what a brand is already sits in its previous collections, so we look at the whole archive and fine-tune a model on it. Then we set guardrails with them, and the guardrails are about what not to do and who not to be. That is where real human taste comes in, and it is iterative.

For someone starting from nothing, this is the best possible way to begin, because you start with a modern workflow. But the founder still has to take the leap of faith about who they want to serve. Technology can build the mood board and the collection. It cannot make that choice.

Sell it before you make it

If you can generate hundreds of designs, don’t manufacture hundreds. Send ten for production and put the other ninety out as a drop, a register-your-interest, an early buy. Then manufacture against the demand you actually see.

This is the SKU-ification of fashion, and if it’s happening on demand everywhere else, why not fashion on demand. Start with small drops, work out what is core to your collection and can be made repeatedly, and build the newness around that core.

What small batches did to 1,200 factories

We have onboarded close to 1,200 manufacturers across the country in two years: knitwear, denim, kidswear, ethnic wear specialists working with crafts specific to their regions. Everything digitised, GST portal, e-invoicing, one click.

When we launched, going from large batch to small batch was a genuine shock to their working rhythms and it created real chaos in their flows. Months later, talking to them again, we found things we hadn’t anticipated. Continuous shipments meant continuous working capital, so their capital requirements came down. And when something went wrong in a batch, it could be corrected in the next one instead of discarding the whole run. Their QC got better.

The actual minimums

The platform is a buffet. We’ve decomposed it into services and agentic systems, so you pick what works. For designs, we expect a brand to take at least 50 from us, because that’s what makes a cohesive collection. For manufacturing we need 150 units per design, which is the point where it works for the brand and the manufacturing partner both.

And low MOQ is a real problem, because nobody wants to work with small volumes. Because we aggregate demand across brands, we have leverage an individual founder doesn’t. The manufacturers understand it too. Today’s small brand is tomorrow’s large retailer, and everyone starts small.

We demoed the platform to a manufacturer in Bengaluru, and the first thing he said afterwards was that he wanted to show his son, because now his son could build a brand on top of the capability he had spent decades building. That is going to be a big push: manufacturers becoming D2C brands themselves. It has already happened in China.

India’s advantage, and the condition attached

For China plus one, India is the strongest alternative, and the reason is simple: the whole value chain is here. We are a large agricultural nation, we make the fabrics, we have the fabric-to-textile ecosystem and the garment manufacturing ecosystem. It is a one-stop solution.

But nothing happened overnight for China, Korea or Japan. It took decades, because manufacturing requires immense patience to reach that precision and throughput consistently. The thing plaguing our industry is that we don’t keep our promises: on timelines, on quality, on labour laws, on environmental laws. Deliver at the right quality, the right time, the right price, and we are on that path.

Since this conversation

Fynd Create has launched in the UK and the GCC. Fynd reports early deployments across fashion brands in India and internationally showing up to 60% improvements in design productivity.

Listen to the full conversation with Farooq Adam on FyndOutWithRagini.