‘Retailers who don’t start using AI today are hopelessly behind’

Retailers not deploying generative AI today to facilitate customer conversations are missing a crucial train. But the scope of AI in retail goes much further, Microsoft expert Manas Bharadwaj knows.

Natural conversations

There is a clear acceleration in the market in terms of using generative AI. ‘The big event of the past year was the launch of GPT-4o by OpenAI. That new, advanced version of Chat GPT took the world by storm. You can now have natural conversations with voice and video, whereas in ChatGPT everything was still text-based,’ Manas Bharadwaj says.

You can connect that language model with simple APIs to your own webshop, for example, so that customers can place orders through natural conversation. The customer simply talks to the webshop. ‘It’s phenomenal! The system understands the context of the search query and gives the right answers. You sit and watch it with your mouth open. It’s not perfect yet, but it’s moving fast.’

‘Conversational commerce’

The application of Gen AI in customer apps is a no-brainer, thinks our interviewee: ‘Customers are starting to expect that. Just as they have been expecting you to have an app since the launch of the iPhone 15 years ago. This new form of conversational commerce will soon become the heartbeat of every brand. If you don’t offer this, you will fall behind. That is the new reality.’

Bharadwaj shows a demo, where a customer shows a hiking shoe to the chatbot in an outdoor shop. Whether these shoes are suitable for his planned hike in the Dolomites next week? The app recognises the shoe, recommends a better choice based on the weather forecast in the region in the planned period, and switches effortlessly from English to Spanish. Impressive. ‘This is what consumers expect: now, not in five years.’

Smart image recognition

Crucial in this application is the development of the intelligent image recognition software AI Vision: ‘It “understands” what an orange is, or an apple. You no longer need to train that model extensively. For example, it sees where there are empty spaces on the shelf. Is the right product in the right place? It also understands what is happening in the shop: how do customers move around the space?’

The application possibilities with smart image recognition are therefore endless. ‘In every shop today, you have smart cameras hanging. You can start looking at customer patterns in the shop. What is the shopping mission they enter the shop with, and how does it change over the course of the day? What is the customer journey? What is the dwell time, for which rayons or products?’ Or think about click-and-collect systems: you can have cameras monitoring the parking. If the customer agrees and gives their number plate, you can see who is coming and get the order ready immediately.

For an autonomous shop called Choice Market in Colorado, equipped with the AiFi technology, Microsoft analysed shopper behaviour using Smart Store Analytics. ‘An example: they sell breakfast burritos there. Those were in the shop until 10am. But the cameras saw that customers kept coming to that department even after 11 or 12am. By expanding availability, they saw a 50% increase in sales for those breakfast burritos. In fact, many people were buying them for lunch as well.’

New generation of robots

Productivity is another angle. A new generation of humanoid robots is on the way, equipped with computer vision and generative AI. One example is Figure, which uses technology from OpenAI, Microsoft and NVDIA to create general purpose humanoid robots.

‘Shop assistants today spend much of their time restocking shop shelves: robots will soon be able to do the same. They can also work in distribution centres. People will no longer have to perform heavy jobs that cause back injuries, for example.’

‘Put this on the agenda’

Bharadwaj sees many more applications, from HR policies over data analytics, promotional planning or deploying digital avatars in customer support. At the seminar ‘AI in Retail’, one of the two supporting programmes at the upcoming RetailDetail Night on 21 November, he will come to explain them.

Is this only for multinationals? What can small and medium-sized Benelux retailers already do with AI? ‘The resources are there. You don’t need a big team of machine learning experts or data scientists. There is an extensive ecosystem of partner companies: in Belgium, for example, you have the Cronos Group. But the heart of the matter is not the technological solutions, but the culture of change and innovation. In many companies, AI is not seen as a priority for the CEO, but it is for the CIO. It is not yet high on the agenda. That has to change. Just start with it. Is it perfect? No, it keeps evolving. Start working on it, you will see it getting better.’

Want to know more? Come to the pre-programme AI in Retail during RetailDetail Night on 21 November, and enjoy a festive plenary programme in the evening with former Walmart top manager Dirk Van den Berghe and Coolblue CEO Pieter Zwart.