Hands-free retail and intelligent retail—How soon, how big?

Reproducibility and prototyping can turn out to be important determinants of the ease and impact of using machine learning algorithms—a user of AI distills some experiences here, writes Pratima H.

His company is not alien to the power of data as it has been helping supply chains with data-driven answers for SKU planning, but Pronam Chatterjee, CEO, BluePi Consulting tells us about the next level of experience and elevation that AI brought to this mix. Let us find out how the company is augmenting human creativity with AI.

Can you tell us more about the role of AI in what you offer? How important, and easy, was it to solve the supply chain issue at the SKU level?

I would say Artificial Intelligence (AI) is the enabler, the key ingredient of our solution. The core of the solution is the ability to understand the deep patterns of data, and to be able to forecast what is most likely to happen in the future. Now if you think of SKUs as independent entities, each one of them has different rhythms, different seasonality, and different factors impacting its performance.

It would be impossible to chart each SKU’s trajectory without AI. Our AI enables the "science" part of the equation or we can say decision-support part of the equation, while for creativity we rely on the designers and product managers. In short our solution augments human creativity with AI.

What role is technology playing in areas like markdowns, store analytics, and omni-channel strategies? How do you nail the 'customized' part in your solutions?

In the last decade with the rise of globalization, Internet, and multibillion-dollar conglomerates, the retail space has expanded both in the supply side and the demand side. Retailers now have thousands or even millions of SKUs, hundreds of stores, and multiple consumer channels.

Business decisions which were based on human intelligence can no longer scale. That is why markdowns, store analytics, and omni-channel strategies all have the best chance of succeeding with the support of technologies like AI and Big Data. The possibilities are endless, starting with business insights to decision automation, to even aiding creative designing.

The key to understanding and building good solutions is to acknowledge that each business and hence each supply chain, has its own nuances, peculiarities, and constraints. These differences are what make different retailers stand out and compete. Therefore, our focus is always to understand their key differentiators and adapt our solution accordingly.

Why did you pick AWS to build your solution? What were the key advantages?

AWS is the market leader and its services, especially in data-related technologies, Machine Learning, and AI are far ahead of its competitors. For us a key enabler was technologies like Amazon SageMaker that reduced our time to market by 5x. Moreover in our experience, the AWS team has always been focused on solving real-world business challenges instead of focusing just on technology, and this really works well with our mindset.

How exactly did the data lake part and Amazon SageMaker part help here?

A data lake forms the foundation of any sizeable data platform. The data lake provided by AWS has the ability to store any type of data, at a massive scale—economically. The different tiers of storage help us build reproducibility into our results and there build business confidence in the outcomes.On top of these core capabilities, we use different analytical engines to support real-time analytics and AI and ML workloads.

Amazon SageMaker is a real AI/ML powerhouse. It has helped us prototype and collaborate much more efficiently within our team. The Amazon SageMaker Autopilot makes it possible for our team to build, train, and tune machine learning algorithms with significant improvements in speed and accuracy.

Is it easy for retailers to embrace AI or automation given the legacy burden, lack of data sets, and contextual data? What major challenges and outcomes have your customers reported so far? Have you incorporated them into your future innovations?

It is definitely not easy for retailers to embrace AI and exactly for the reasons mentioned above. But most of these challenges can be addressed technically or sometimes with human intervention.

However, in our opinion, the single most important key to success for any retailer is to have a consensus amongst the senior leaders to make the transformation successful. That is why a part of our initial engagement with customers is to help them analyze and ascertain the positive results they can expect by adopting AI. Our approach is to help our customers arrive at a potential ROI for their specific use case.

Even before the start of the project, it is possible to demonstrate the potential impact on the top line or bottom line by adopting AI in the organization. This exercise lays a solid foundation and also helps our customers understand the process and outcomes.

Would the industry be moving towards more hands-free retail, click-and-go shopping, scan-and-pay checkouts, and robotic assistance as we step out of lockdowns? How will you adapt, and help retailers to adapt, to future possibilities?

Ensuring safety for customers and employees is most critical that is why click-and-go shopping, hands-free retail, and robotic assistance are expected to see an increase in adoption. We are also exploring possibilities with our clients to adapt their store layouts and modify the assortment mix, which will help them quickly pick these capabilities.

Post lockdown, one of the immediate areas of concern for most non-essential item retailers would be to optimize their inventory through liquidation, markdowns, and discounts. The idea, in essence, would be to free up as much working capital as possible and adapt to customer demands. We are working closely with our customers to help them prepare a business resumption strategy.

What next can we see come out of your think-tanks? What technology area excites you most? Will you continue to collaborate with players like AWS?

There is growing appreciation and demand for data-driven business transformation, moving on from digital transformation initiatives. The driving force behind digital transformation was mostly technology advancements, but a data-driven business transformation is mostly being fueled by organizations trying to achieve strategic advantage.

The areas that are most interesting to us are the new business applications and adoption paradigms for AI that are continuously evolving across industries. I think we are still at the nascent stages of the AI revolution and the real impact is still to come. We continue to innovate applying in new and creative ways to solve real business problems. An example that comes to mind is that of using Convolutional Neural Networks (CNN) to analogous products, which is a key step inforecasting demand for new products.