Mellowain was a start-up which aimed to reduce queues in the stores. We were making a shopping cart to which a screen was attached, so whenever you add a product it got scanned in the cart itself. This reduced the time of scanning near the counter and hence eliminated the queues.

1) Display Upper time - After the customers are done with picking up things from the store, i.e. when they are standing in a queue for check out, we display the time/upper time in which they will see check out ( or the waiting time). This will reduce the anxiety of customers. Boon for Mellowain- by using Mellowain Trolley the time near counter is negligible.

2) Traffic near each counter- I also came up with an idea where we can show the traffic in each queue so that customers can choose the smallest queue or number of open queue. This would be done using thermal imaging, hence would be added in the later stage.

Later, I worked on benefits the ‘Stores’ would have if they keep our cart in their stores.

1) First and the most obvious was if there are no queues it will directly increase the revenue of the store.

2) Display total savings ( discounts on products) so that the customer is always aware of their savings, which would give good feedback to their customers of the benefits of shopping at their stores.

3) In-Aisle recommendation i.e. provide them with recommendations and special offers on products of the same aisle, rather than displaying ads of products of a faraway aisle. This was implemented using ‘content-based recommendation’ and with the database of the aisle. Search option (aisle providing aisle with no of the product ); Recommend New Product in the aisle / Special offers per aisle.

4) Search option (aisle providing aisle with no of the product ); Recommend New Product in the aisle / Special offers per aisle.

Our product required users to login(for building recommendation system), which is always a painful task as these days one user has a lot of usernames and passwords. It is hard to remember which username and password were meant for which site. To reduce this confusion, I worked on a feature which helped to solve this problem. Basically, when you type a username, it will give you confirm whether the entered username exists or not if it exists you are directed to password row. This was implemented using ‘Two-Part Login System Algorithm’

When I came to Mellowain I was first introduced to a book ‘Hooked: How to Build Habit-Forming Products’ by ‘Nir Eyal’. My first work there was to come up with solutions to all the exercises in the book, the solutions were later used in the actual product.

According to Nir Eyal, there are four important aspects that are very important in building product- Trigger, Action, Variable Reward and Investment.

The Triggers of our Product were- it Saves time, Frustration of standing, Painful task of scanning each and every product is eliminated. The book States that -The Trigger is useless if the user doesn’t take action. To initiate action doing must be easier than thinking.

The Actions of our products were – as we are eliminating queue most people would like to use it, at a later stage we have planned to put local languages.

The Rewards for our products were- Store pays for a cart if the bill reaches y amount. Otherwise, the customer pays for it, Discount / free to use once a week, rewards uniform throughout all the stores where a cart is used, Co-shoppers/group/family ( rewards distributed), Rewards based on the history of purchase, Free Parking (no- cash), Cab fare-free till a distance (non- cash).

I was also designing UI in Mellowain. The kit that was used for programming the screens was by NVIDIA. Since the programming was done in Ubuntu and the environment for it is supported by Qt, I used Qt to design UI screens

The product heavily relied on image processing, as I had worked in this field in my final year project. My previous knowledge not only helped me in understanding the product better but I could also give valuable suggestions, like- I had worked on MATLAB for image processing, but it is very slow and doesn’t give accurate results. Rather using OpenCV which is not only open software but also give fast and precise results