Streamoid builds on AI-aided image search

By Anusha Ashwin There is a multi-fold increase in e-commerce websites, globally. According to analysts, popular website creators worldwide develop a minimum of two e-commerce websites per month. It is but natural that companies and professionals are innovating in building highly user-friendly e-commerce websites, integrating features that enhance images-based search using Artificial Intelligence (AI). One such company is US and Bengaluru, India-based Streamoid Technologies Inc. It has created a technology to recognize an image and then search a large visual database to get related content online.

Streamoid Technologies is a visual-based search and discovery technology provider that came into existence with a mission to “Put Images to useful Actions”. Streamoid offers a set of vision-based tools that provide newer ways for people and businesses to find, explore, collect and share image-based content.

Founded by former Senior Director of Nvidia, Sridhar Manthani, Stanford Researcher Haricharan Lakshman and Rajesh Kumar, Streamoid is a company that seeks to find solutions for fashion retailers using its computer vision technology and AI tuned by fashionistas and stylists. Incorporated in March 2013 in Palo Alto, USA, with the Indian subsidiary, Streamoid Technologies is slowly attempting to revolutionize web search, where browsing and typing key words would seem like a thing of the past.

The piQ portfolio

“It was quite apparent that visual content is going to be pervasive across different apps and websites. With this proprietary research in hand, we set about thinking how information available in images can be used to improve the customer experience,” recounts Co-Founder Sridhar Manthani elaborating on how the seed of thought to set up the company was sowed. “We decided to focus on publishing and fashion e-commerce. We built the piQto image search engine that operates in publishing. The piQit Fashion platform is the evolution of this. We were on the cusp of the e-commerce revolution in India, and believed that fashion is visual, and we can use our technology to help online retailers and shoppers,” he said.

Initially, when the founders came together they developed their first product that targeted media houses with large databases. They developed an app called PiQto for Publishers, which allowed a user to click a picture of an item and the app would then fetch all the content in the media house’s data base related to the item. piQto was developed as an image recognition tool that converts the mobile camera into a visual search engine. It enables the app user to click on print ads/images in partner publications and see related digital content like video’s, product catalog, picture gallery or take a quiz/poll or buy the product online.

Explaining about their first product Manthani says, “piQto is our image search engine and is the initial proof of concept for our technology being used in our publishing product. It allows any image to be connected to additional related, whether existing on a website or appearing in offline medium like newspapers/magazines or billboards and find an immediate match extremely quickly and efficiently. This ultra-fast lightweight algorithm allows all publishers such as newspapers and offline or online magazines to monetize their images and bridge the gap between the offline world and the online world.”

PiQto for Publishers may be an off-beat, advanced, and innovative product for the publishing industry, but the founders realized that there were very few takers for their innovation. “The publishing industry definitely sees the value in the piQto image search engine but is a bit slow to adopt new technology to their existing business. Once the publishing industry realizes they need to focus on monetizing images and attach commerce to content, we will instantly reap the rewards,” says Manthani.

Obviously with a good futuristic technology in hand, the founders were smart to immediately parallel their focus on the ever evolving e-commerce industry. Rajesh Kumar, the chief engineer of the company, along with Manthani and Harish Lakshman created piQit Fashion catering to the $1 trillion global fashion industry. “It is not a search engine but an entire platform that we think is going to dramatically change the level of personalization. It comprises a consumer app in addition to integrated set of tools for retailers to increase their conversions. Users can use the piQit app to discover new fashion, share with friends, try on outfits virtually, and store all their fashion purchases and interests in a single location,” explains Kumar.

The piQit Click&Find enables users to discover products that are inspired by real life. Shoppers can click or upload images from print catalogs, magazines, Pintrest or even street fashion and find similar items in the retailer inventory. piQit Similar empowers fashion e-commerce retailers to provide similar product recommendations based on color and pattern. Streamoid’s visual search engine matches images being browsed and picks similar products from large databases and serves it with speed and accuracy. Streamoid’s apps have the capability of yielding ultra fast visual search results with responses under 0.1 seconds. piQit Outfitter enables online fashion retailers to inspire purchase by making real time outfit recommendations around the product selected, replacing the need for manual curation.

“We allow users to save all the items they have ever purchased in the piQit Cloud closet, which stays with them wherever they go. On top of this, they can place outfits on pictures of themselves to see how they look, they can chat with a stylist who has access to all their preferences and past outfits, and they can take a picture of an item and buy something similar. This same piQit user, when they are browsing a fashion website, will see recommendations for clothes they should buy from that website to make a great outfit with what they already have. They will also be able to chat with the virtual stylist about clothes/sample outfits being offered on the website. Again, any items they buy from the site will be automatically saved to their piQit closet. So, the user is able to go from site to site and have all their items saved to a single location, and have a personal stylist constantly following them and helping them make styling decisions wherever they go. This virtual stylist/shopping assistant sits on the retailer’s site and drastically improves the purchase confidence for the customers, and helps ecommerce retailers to cross-sell and increase the Average Order Value of each transaction,” details Kumar.

The first breakthrough for the company came when they won the deal with Trendin.com, an online retail initiative from Madura Fashion and Lifestyle (an Aditya Birla Company). Trendin offers branded apparel from popular brands like Louis Philippe, Van Heusen, Allen Solly, People and Peter England. “Trendin was able to drive enough traffic to the site. But, the product views on the site were not translating into purchases. The Click Through Rate for similar recommendations was always less than 5% and addition to carts was even lower. Also, being a multi-branded fashion apparel company, customers were finding it difficult to navigate the site looking for product choices. The previous recommendation algorithm was based on ‘collaborative filtering’, which used, user’s past purchases, shopping cart additions, social history, and what other customers have viewed and purchased. This did not work well for Trendin site. That is when Streamoid stepped in. Trendin partnered with Streamoid Technologies in January 2014 to simulate customers’ shopping behavior. Streamoid’s visual technology piQit was integrated into the Trendin website to drive recommendation based on visual attributes like color, pattern, and style. A product page view also presented visually similar products across the store. This emulated the typical behavior of an Indian customer in offline retail shops. This increased Click Through Rate and resulted in higher conversions,” recollects Manthani.

Patents et al

Trendin’s success gave the founders the confidence to freeze deals with five other top fashion e-retailers and they are positive on signing more. The founders believe that startups cannot go chasing copycats. It was quite natural for them to file for IPR and currently Streamoid has three patents in various stages of approval. “We don’t think patents are very important as an offensive tool. It is more important for us to focus on offering our customers the best product and experience and innovate faster than anyone else. Patents could be useful for defensive purposes but inherently focus on the past,” says Manthani.

The lean way to scale-up

Streamoid is 14-member strong and was bootstrapped for a year until they received $1 million in US angel funding in 2015. Not disclosing the name of the investors, Manthani said, “At this point we are focused on growing business and rolling out new features to our piQit platform. We are always searching for strategic investor-partners who can help grow the business using their industry expertise. The goal is for our revenues to maintain the company’s day-to-day operations, and any external capital we take would be used to power growth and R&D. Currently, we are noticing a serious overuse of investor capital in many companies, whose overhead expenses and customer acquisition costs balloon at the investor’s expense. We believe in the concept of a ‘lean machine’ and utilize capital efficiently, i.e. getting the most bang for the buck. We are generating revenues today and we aim to be profitable soon.”

Having an office in the US would mean that the country teaches them on leading indicators on trends in technology and product/market fit. The team imbibes ideas by talking to customers, investors and influencers who are based in the US/Europe to be proactive in making better products and react faster to market. It is obvious to Manthani and his team that Artificial Super Intelligence is the future. As AI research and hardware performance improves, AI will become more and more important in their product development. Manthani believes that in the near future, Fashion AI will absolutely transform customer discovery and Streamoid is positioned well to be in line with the technology’s transformation.