Bengaluru, Karnataka, INDIA: For somebody pushing 40, getting into a gym for the primary time could be a nerve-wracking experience.
As this author realized, even before self-doubts crept in about their possibilities of surviving an hour on the gym, the larger looming query was what to wear and never look misplaced.
With absolutely no idea what they were searching for, this author turned to an AI shopping assistant by Myntra, India’s biggest online fashion retailer, and typed, “I’m searching for clothes I can wear to work out within the gym.”
Surprisingly, the AI assistant understood exactly what this author needed and got here up with jerseys that may wick off sweat, compression t-shirts, self-proclaimed comfortable trackpants that wouldn’t restrict movement, shoes that might make you run higher, fitness bands and all types of gear a newbie couldn’t have imagined they wanted or needed.
With the shopping cart full and the wallet significantly empty, this author was ready for a brand new starting.
What the AI assistant did – convert an abstract user query into actionable results – is game changing for the style industry. Conventional search works best with specific keywords – a blue t-shirt from a selected brand, say.
It goes a couple of steps beyond conventional search. It uses generative AI to answer more open-ended questions like what to wear for a selected festival or a cricket match and even the trending fashion in a city.
“That is big,” said Arit Mondal, director of product management at Myntra, “Why? Because, that is the primary time we now have an answer, which is solving the unsolved ‘search’ problem in the style, beauty and lifestyle industry. And it’s live for purchasers at scale.”
Because the starting of online fashion retail, looking for products has been very much like looking for every other piece of data online. You are trying a set of keywords and keep refining your search with different keywords and preset filters.
A seek for a branded, blue t-shirt works well since the keywords are already a part of the product catalog.
But that’s not at all times how people shop in the true world. Some shoppers only have a vague idea what they need – as an example, clothes for an upcoming vacation or a rock concert.
The standard approach to searching by keywords fails spectacularly in relation to the second sort of customer because the search strings they use should not retrievable directly from the data stored within the product catalog.
Until now.
When generative AI – built on large language models (LLMs) that synthesize vast troves of information to generate, text, images and more – first made news last yr, the team at Myntra quickly began fascinated by how they may leverage it to reinforce customer experiences.
When Myntra organized a hackathon in February this yr, a bunch of engineers from the corporate’s search team decided to make use of Azure OpenAI Service to resolve the abstract search problem and unshackle users from the cuffs of keywords.
They were pleasantly surprised to see how ChatGPT, the generative AI service available through Azure OpenAI Service, could synthesize natural language prompts. They asked ChatGPT in regards to the look of an actor from a recent movie and it could tell it consisted of a bomber jacket, gloves and aviator sunglasses.
“And that is the data that Myntra’s existing catalog didn’t have,” said Swapnil Chaudhari, an engineering manager at Myntra.
Over two days, his team took over a conference room and kept trying latest prompts – text that generative AI could understand – to see what results they got. This was latest territory – and so they didn’t know the way far they may push.
“We were surprised to see the outcomes. It was in a position to answer questions like clothes to wear for regional festivals like Pongal and Onam,” said Pragna Kanchana, a frontend engineer at Myntra.
On a whim, she tried to look in Hindi with , which in English translates into winter clothes. And it understood it!
The team then got access to Azure OpenAI Service’s playground that allow them do way more than was possible with ChatGPT alone.
“Leveraging Azure OpenAI Service, we were in a position to plug in numerous large language models in the identical prompt and determine which model worked best for our use case. So, we had plenty of freedom to match and select the precise model,” explained Santanu Kanchada, a backend engineer within the search engineering team.
The team knew they were on to something big. They wrote the code in a day, and inside two days they’d a working prototype of a brand new feature that enabled users to look with natural language.
“If it weren’t for GPT models, we’d need to first retrain the model using Myntra’s catalog after which wait and check the outcomes with our expectations. However the pre-trained models already available with Azure OpenAI Service were already performing quite well,” added Chaudhari.
Over the subsequent five weeks, multiple teams across engineering and product development fine-tuned each the backend and the user interface for the AI shopping assistant.
“Myntra’s systems are on Azure and deploying Azure OpenAI Service was as seamless as deploying one other server and it gave us a secure way of using generative AI,” explained Vindhya Priya Shanmugam, director of engineering at Myntra.
Post the hackathon, the search engineering team kept refining the prompts to get useful outcomes for users. Certainly one of the issues, as an example, was find out how to be certain that the response to a user’s query resulted in clothes for under the gender the user is searching for.
Within the weeks resulting in the launch, they trained the system on Myntra’s catalog and added guardrails so the outcomes were limited to the catalog.
The AI shopping assistant was launched on the Myntra app in late May, just in time for certainly one of their biggest marquee events, End of Reason Sale (EORS). It included sample prompts that gave users an idea of how they may use conversational language fairly than keywords.
Since then, Myntra has already seen search queries broaden, offering latest opportunities for product discovery. As an example, when someone searches for garments they’ll wear to a beach, not only beach wear but in addition accessories like hats, sunglasses and footwear pop-up.
It has been phenomenal for Myntra.
“Users who shop using the AI shopping assistant are thrice more prone to find yourself making a purchase order,” said Mondal. “Since it also helps users discover an entire look from multiple categories of products, we’re seeing that on average they add products from 16 percent more categories than usual.”
While this author’s fitness transformation journey remains to be questionable, multiple teams at Myntra are already constructing latest features based on generative AI.
Certainly one of them will allow users to decide on different categories of products – tops, bottoms and accessories, for instance – and see how they appear together in an outfit. Myntra plans to further enhance it by introducing voice search and supply personalized results. Also they are how they’ll use generative AI to assist the client support teams.
.