Contrary to popular trends, I’m one of those who really doesn’t enjoy shopping. Just the thought of sifting through piles of clothes and accessories is extremely taxing for me. I imagined I’d feel differently about online shopping, but I enjoy that experience of virtual shopping even lesser! My deal is simple – I need to find what I need within 5 minutes of strolling into a shop or browsing through a site/app. My entire transaction is usually completed in 15 minutes.

So when I spoke to Sanjoy Roy, cofounder of AskSid.ai, I was delighted to find out that there is technology being built for irascible shoppers like me. Roy, who previously worked at Mindtree along with a colleague Dinesh Sharma, started AskSid.ai in 2017, after his wife Dolly struggled to find a good pair of leggings in the Netherlands. Dolly, who was pregnant at the time, was looking for a pair of leggings that would be breathable and expandable. To her dismay, a visit to a popular online shopping site showed her nearly 400 options. Feeling completely confused, Dolly decided to just visit a store instead and she got exactly what she wanted in 20 mins. “I was left wondering how this experience and ease can be replicated through an app or site. Stores clearly sit on a lot of data of their inventory and shoppers preferences, so how can this information be made readily available digitally – this was what Dinesh and I were intent on solving.”

Today, AskSid.ai is one of the few platforms that’s a full stack solution for the retail and consumer goods, present in more than 23 countries and in 16 languages. The solutions provided by AskSid are aimed at three main activities:

• Help retail companies save money
• Improve conversion rates and improve customer service
• Provide market insights for a competitive edge

This is being accomplished by a range of components on the AI platform like AskSid, which is the flagship conversational platform; ReplaySid for tracking consumer conversations; SwitchSid for a smooth handover from the AI platform to a human agent; MeasureSid for tracking analytics; MonitorSid – a dashboard for providing business KPIs for brands; IntelliSid that hosts the AI domain metadata and TrainSid, a cognitive platform for algo testing and data enrichment. The AskSid Canonical Data Store is another important feature where raw data is used by algorithms for data enrichment.

Not Just Conversational AI; Measuring Real Impact for Brands Worldover

Roy is a huge believer in the potential of AI addressing some of the retail industry’s most pressing challenges. Despite being headquartered in Bangalore, AskSid deals with retail and CPG companies across the world ranging from clothing, wellness to home furnishing and dairy. The range in clients has helped Roy and his data team develop custom solutions for unique problem statements. For instance, one of AskSid’s earliest clients was an Austrian apparel brand – considered among the leaders in Europe – had a large volume of queries in assisted clothing, especially in the lingerie section. Another client is a world leader in performance coatings revealed some interesting insights. Roy says, “In India, paint jobs are outsourced to contractors, who also usually decide the brand and even the colour of paints used in buildings. Abroad, this is a largely DIY market. This particular client would receive a lot of queries from B2C as well as the usual B2B clients. While they had all the product inventory available, it was a tedious task to find the right kind of information based on the incoming query. Our solution helped aggregate all these data points to answer queries much faster.”

He also recalls working with an Indian company that is a segment leader in producing healthcare and wellness products, which would receive queries from consumers about the side effects of diabetics and heart patients using some of their health supplements. The company’s marketing division was unaware of such consumer demands, and this information funnel helped the company build targeted marketing campaigns for their products.

“Every 100 conversations through AskSid.ai would lead to atleast 11 online orders in a period of 7 days."

Managing Diverse Data, Intelligent Machines and Bias

Being present in over 20 countries and training its algorithms in 16 languages too has brought its own set of learnings, says Roy. Not only does every country has its own unique language, but these languages have their own style and syntax. From a machine learning perspective, it means more retraining and tweaking algorithms to understand customers more intuitively and provide the best in-store experience.

The advantage of working with world-class retail companies also means the AI’s capabilities enhance due to the availability of large and diverse datasets. “The AI is advanced and can answer complex queries and this is due to the range of data coming in. We take a company’s product catalogues, PDFs and other forms of data to crunch it into bite sized nuggets, and the AI is able to extricate valuable insights from these vast amounts of data in very little time.”

Dealing with multiple markets and customer profiles also means being wary of data bias. This is a very real concern in the AI space today, admits Roy, and says his team of data scientists are continuously working on mitigating bias in their algorithms. “We take the source algorithm and train it for specific use cases. We know the bias will creep in, and our data teams constantly monitor the algorithms for bias.”

COVID19 may have spelt doom for many businesses, but some others have survived and even thrived. AskSid is one of them. Roy said they witnessed a nearly 4x jump in income through online sales and they have diversified their virtual advisory services. In addition, they have also rolled out to more international markets in the past six months. The plan remains to stay on track with these positive growth rates and expand into more markets for a bigger online retail share. 

Jcomp Freepik