Athiya Rastogi, CEO and co-founder of Toronto-based SnapWrite, which automates online second-hand retail processes, is dressed in preloved items: 'It's amazing to see how the world has opened up to resale.'Don Campbell
As a student studying statistics and machine learning at University of Toronto, Athiya Rastogi indulged her passion for fashion and earned extra cash by running an online second-hand clothing store.
She soon determined most of her store operations could be automated, and in 2022, she co-founded SnapWrite, with her brother, Aryaman, as a way to streamline ecommerce for other second-hand store owners.
“Sellers of preloved items are particularly burdened when it comes to digitalizing inventory because everything in their shop is one-of-a-kind with no digital history,” says Ms. Rastogi, whose early career included automating her family’s global homeware business and conducting data research for Unilever and other brands.
“If you’re putting out 2,000 items a month, who’s going to sit down and write up all those descriptions? It just doesn’t make any sense,” says Ms. Rastogi.
The AI revolution in resale shopping
Using SnapWrite’s AI-driven image-comprehension tool, clients can upload a photo of an item, and in seconds, a detailed product web page is created, listing size, brand, colour, materials, condition and up to 15 other attributes.
The new web page, complete with title, photo, on-trend description and metadata – replete with predictive keywords for search engine optimization – seamlessly integrates with the client’s store on Shopify, Magento, LightSpeed and other online retail platforms.
In addition, SnapWrite users can automate consignor tasks unique to recommerce, such as post-sale commission processes.
The Toronto-based startup’s clients, 50 stores and counting, have digitized 120,000 items, saving thousands of work hours, says Ms. Rastogi, including for U.K.-based children’s clothing and toy retailer Bon-Tot.
We are not just superimposing clothing on an avatar. When you twist and turn you can see on your screen exactly how the garment looks on you.
— Sulochana Karthik, co-founder and chief technology officer, Haute-U AR
To capture the nuances in language and product features, SnapWrite moved operations to partner with customers and created datasets and corpuses from their data, which allowed the company to create their own models.
Ms. Rastogi adds that in the midst of these advances, data privacy remains a paramount concern, ensuring that all client information is handled with the highest level of security.
“The user has complete transparency of where their data goes and what happens with it, right from the time data is inputted by the user in the platform,” she says. “Each user’s data is stored in an isolated fashion to protect against any data breach.”
Resales automated
SnapWrite isn’t alone in using AI to generate product descriptions from images. Amazon and eBay have launched AI-driven product description tools, while Chat GPT-4 can apply language-reasoning skills to images.
Ms. Rastogi says SnapWrite, which has received funding from big-league venture capitalists, including FounderFuel-Real Ventures, Inovia Capital and Panache Ventures, as well as a few early clients, goes further.
“We’ve designed a system that takes resellers’ entire process from the beginning to the end to help them get items sold online,” she explains.
Innovations like SnapWrite reflect an accelerating recommerce trend. According to Statista, the global resale market is growing five times faster than traditional retail and is expected to climb from $197-billion in 2023 to $300-billion by 2026.
“It’s amazing to see how the world has opened up to resale,” says Ms. Rastogi, who ultimately wants “everyone to be able to do it easily with a click of a button.”
Try it on
A love of fashion also propelled a former data software developer in the wealth management domain to transition her expertise into pioneering a revolutionary app that will allow online shoppers to try on clothes in an augmented-reality environment.
“We are not just superimposing clothing on an avatar,” says Sulochana Karthik, referring to existing try-on apps. “When you twist and turn you can see on your screen exactly how the garment looks on you.”
Sulochana Karthik, co-founder and chief technology officer of Haute-U AR Technologies.Supplied
Ms. Karthik, co-founder and chief technology officer of Toronto-based startup, Haute-U AR Technologies, aims to make online shopping as engaging and precise as visiting a physical store as a way to boost sales and gain customer loyalty but also to reduce sky-high product returns – “a disease aggressively attacking profit margins,” according to Shopify, which calculates that one in four purchases are sent back in costly reverse logistics.
To a large extent, Haute-U AR’s “try-before-you-buy” platform relies on AI algorithms that recognize people’s faces and bodies to drive augmented-reality video communication that allows shoppers to see themselves “wearing” a product – like they can do in physical stores, which see just 10 per cent rebounded goods.
The platform will also include “a generative AI-powered recommendation engine,” able to suggest products based on a user’s face and body type, says Ms. Karthik.
The team faced significant technical challenges, including securing adequate resources of AI GPUs (graphics-processing units) due to the increasing demand for Cloud services and integrating AI technology into the app, she says.
“To achieve an exact fit for real-time users in AR is complex,” she explains. “We are collaborating with game developers to refine the technology, which also allows us to process user data, such as image uploads, without the need to store it, ensuring greater data privacy.”
Positive feedback on the testing phase was encouraging, says Ms. Karthik, and validated the app’s intuitive design and AI-based recommendation system, with one user describing the experience as like “looking into a magic mirror.”
Dark-skinned, green-eyed and Rubinesque? Haute-U AR’s personal AI stylist will whiz around the world’s online fashion market, choosing perfectly sized and suitably hued clothes, accessories and even cosmetics to complement those features, says Ms. Karthik.
In her aptly called “Fashion Playground” ecosystem, set to launch next year, users can try any number of products – hundreds, even – virtually instantly, including products they wouldn’t normally try in a store, advancing the platform’s strategy as a product-discovery tool.
Haute-U AR, which this year received research assistance from Humber College and participated in Elevate’s Accelerator program, sees Fashion Playground, “as a way to provide a broader scope than other AR try-on developers,” says Ms. Karthik, including 3DLOOK, in New York, Zyler in the U.K. and Banuba in Eastern Europe.
“We are actively exploring a partnership with Google to enhance and expand our AR technology,” she says. “As we continue to refine and strengthen our capabilities, we aim to position ourselves to collaborate and compete with industry leaders like Amazon in the AR space.”
The price is right
As startups evolve, their missions often shift. In 2019, University of Saskatchewan research professor Melanie Morrison founded BetterCart Analytics as a grocery price-comparison tool to help consumers, including herself as a single mother of three, save money by finding the lowest-priced items.
But after building an AI-powered system capable of importing 38 million product records weekly and amassing one of Canada’s largest food-pricing databases comprising billions of product records – Dr. Morrison, CEO of the startup, and her team pivoted two years ago, transforming BetterCart from a consumer service to a business service.
Melanie Morrison is CEO and founder of Saskatoon-based, AI-driven BetterCart Analytics, which helps indie-grocers and startup food and beverage makers make informed pricing decisions.Liam Richards/The Globe and Mail
Expanded market-based research had “indicated that small- and medium-sized companies were feeling under-resourced when it came to technology that could drive their pricing strategies,” says Dr. Morrison.
The company currently provides pricing analytics to independent grocers and startup food and beverage manufacturers to help them compete with major retailers like Walmart and Costco and brand giants like McCain and Maple Leaf.
With 10 clients that collectively run 15,000 stores, “there’s a chance for us to reduce pricing en masse for consumers, as well as really strengthen those indie businesses,” says Dr. Morrison.
“Our competitive intelligence lets them capture when [a competitor] drops their price, so they can drop theirs as well, and fight the good fight.”
As Dr. Morrison explains, an indie-grocery store manager planning to introduce a new brand of yogurt at $5.99 can see, using BetterCart, a nearby Loblaw is selling same-size yogurt tubs at $7.99.
The store manager could promote this price difference to attract more customers and help the new brand increase sales. Similarly, if the new yogurt was $10.99, a price-point decision could be made: Is it possible to price it lower?
“The idea is to break up the stronghold held by large brands and diversify our food supply,” says Dr. Morrison, who also says, “BetterCart’s insights are so powerful they can double retailer and manufacturer profit margins for products they are tracking.”
Pricing is the number one driver of consumer purchasing behaviour and roughly 75 per cent of independent grocers and food manufacturers operate without any price analytics, says Dr. Morrison, accounting for recent funding from Canadian Food Innovation Network and Saskatchewan Polytechnic.
The latter, totalling $675,000, will enhance the startup’s machine-learning techniques.
“We’re working on AI-driven predictive analytics,” says Dr. Morrison. Using its vast historical price database of more than five billion product records, she says, BetterCart will be able to predict future product prices, to enhance strategics for additional profit margin gain.
BetterCart recently added its first American client as part of an aggressive expansion strategy.
Should the company’s scale-up succeed, it will buck the Canadian trend of lagging behind other developed economies in adopting innovative technologies.
Canada’s low positioning is a result of a lack of tax incentives and government funding, among other things, according to Ottawa-based Centre for Canadian Innovation and Competitiveness.
To achieve the same level of government expenditure of research and development as a share of GDP as the United States, for example, Canada would need to increase spending by 144 per cent.
This month, however, the Canadian government announced a $2-billion AI plan, which earmarks $300-million for AI compute costs for businesses.
In the meantime, companies like BetterCart, SnapWrite and Haute-U AR will forge ahead on their own initiative, because, says Ms. Rastogi, the work needs to be done.
“There’s no software in the Canadian market offering these solutions in the resale sector, so we decided to do it,” she says.