Entrepreneurs have no respect for the status quo; they are quick to pick up on things and confident in applying learned skills and knowledge.
Entrepreneurs have no respect for the status quo; they are quick to pick up on things and confident in applying learned skills and knowledge. Take the case of Atul Rai, the co-founder & CEO of Staqu Technologies. This Gurgaon-based Artificial Intelligence start-up has developed an innovative image-to-image matching system, hence simplifying image search, automated meta-tag generation and real-time recommendation for e-commerce companies.
Atul is not new to the start-up ecosystem. He has served as the founding member and software developer for Sysfield Technology, where he successfully designed OCR (optical character recognition) based Android app and M-Coupon, another Android-based app, used for advertising discounts by sending coupons, instead of SMS on Android phones. He followed the feat by operating as the founding member and Lead Researcher at Cube26, where he worked on the Blink to Capture technology, used for Blink Play, an eye blink based selfie app for Panasonic phones.
His deep-rooted commitment towards exploring Artificial Intelligence inspired him to start Staqu in 2015, along with Anurag Saini, Chaitan Rexwal and Pankaj Sharma. Their venture endeavours to automate image to text and text to image discovery, hence disrupting the fashion e-commerce, an industry predominantly dependent on enticing product images, says Atul.
Last year brought success for Staqu in two respects—B2B and B2C. “We have collaborated with several e-commerce portals, such as Paytm, Yepme, Roposo, Tradeindia, Microsec Finance, Faballey, E-poolers. Furthermore, we worked in close association with OEMs like Karbonn, for the launch of its brand new phone, FashionEye that is powered by Staqu’s AI,” says Atul.
Furthermore, Staqu has launched a smart fashion app, Fashin to act as the personal stylist and guide for fashion enthusiasts. This app is built with superior AI capabilities to extract fashion from a YouTube video, enable fashion enthusiasts to search for their favourite clothes simply by uploading images, instead of having to describe the elaborate patterns and fabrics and also compare the prices from different e-commerce sites.
Information is key
According to Atul, the Internet consists of 70% of image data and there are only a few companies which are utilising it to solve some of the known problems of search, discovery, recommendation etc. “Staqu started with the thought of solving real world problems, by leveraging the unique capabilities and applications of AI in image domain amalgamated with textual analysis. At the same time, the economy was witnessing a boom in the e-commerce sector. Since the e-commerce sector heavily depended on images, we conceived an invigorating opportunity in optimising the image based AI in the domain of fashion e-commerce,” he informs.
With its prime focus on AI research, image processing, deep learning, computer vision and NLP, Staqu has produced VGrep API suite, fundamentally offering a virtual search engine and a hybrid recommendation engine.
Talking about funding, the co-founder explains: “The first step for us was to emerge victorious in the Smart Champ Challenge, seen as the most promising start-up in the region round, that is in North India. Staqu was picked amongst the top four start-ups by IBM’s Global Entrepreneur Programme in 2015.
The feat empowered Staqu to successfully raise funds from the Indian Angel Network in an investment round led by Ajay Gupta, Bikky Khosla and Neeraj Singal. The investors perceived value in the products developed by the company, which can be leveraged to increase the sales of fashion e-commerce portals, provide a seamless user experience and help OEMs include an additional revenue stream.”
Going forward, Rai says, “We will be coming up with superior and more robust AI which we can use to develop cross-domain recommendation engine and wrap up our AI with different use cases. Furthermore, in terms of technology, we shall be interested in a light weighted neural network, in order to make artificial intelligence remotely accessible and not strictly limited to the internet accessibility.”