Why retail should embrace AI for truly personalised CX

Hiring interior designers, fashion stylists, or bespoke travel agents to level up our lifestyle is a luxury that’s out of reach for the vast majority. Or it has been until recently, at least. Now, artificial intelligence (AI) advancements mean that these services are about to expand beyond the sole preserve of the rich and famous. AI-powered Conversational User Interfaces (CUIs), underpinned by large language models (LLMs) are set to truly democratise the provision of bespoke services. Savvy retailers can gain a first-mover advantage and offer new products to customers on a large scale by using this technology. It’s set to supercharge our shopping experiences, introducing intelligent virtual assistants that sound and act like their human counterparts. And where such assistants might have existed in some form in the past, they would have been programmed with rules and logical paths. Now, they are trained on vast amounts of information, and tap into proprietary data, like a product catalogue and an individual customer’s purchase history. To understand the value of a virtual assistant, it’s worth noting that the vast majority of consumers currently interact with an online retailer through graphical user interfaces (GUIs), usually in the form of browsing apps and websites. These kinds of e-commerce and m-commerce experiences are more convenient than having to visit a physical store, but they’re passive and rely on a potential customer’s imagination to do a lot of heavy lifting. While an app user might be able to get an idea of what a particular product looks like and perhaps read a brief description, they still have to guess how it might look in their home, garden or on their person by picturing it in their mind’s eye. Taking home furnishing as an example, some apps do offer augmented reality (AR) product visualisation (more on this later), but by and large, for big-ticket items in particular, it can feel high-risk for customers to make a purchase based on this level of guesswork. There’s often a reluctance to buy on this basis, leading to suboptimal conversion rates for retailers. The value of the virtual assistant Here’s where virtual assistants can add value. They can provide the additional context needed to get that purchase over the line by reintroducing a conversational ‘back and forth’ element to the interaction (understanding specific and/or complex requirements, curating relevant options and making expert recommendations) that would previously have relied on a one-to-one human conversation. While AR can place products in our existing space, the supercharged virtual assistant of tomorrow will help us completely reimagine our space, not only adding new, virtual objects, but also visualising what the space would look like with existing objects removed. Along with recolouring the walls, adding lighting and such, we can really get a feel for what a space could look like. And the best part? They can do this at scale, at a much lower cost, and at a time and place of the prospect’s choosing – adding value for retailers and consumers alike. This is conversational commerce in action. How does conversational commerce work? You could be forgiven for hearing the terms ‘conversational commerce’ and ‘large language models’ and thinking: “Hang on, chatbots are everywhere already.” You’d be right. But conversational commerce actually refers to the virtual assistants powered by cutting-edge LLMs that will be far more adaptive and multi-dimensional than previous iterations programmed with rules and logical paths. ‘Language’ is also used in the broadest sense in relation to an LLM. While it might mean text, it could just as easily be a model trained on images, audio, video, voice recognition, or even proprietary product data. And the latest iterations are trained on all of these data sources together, with all inputs and outputs processed by the same neural network. This vastly increases the scope of their capabilities. Not only does it give consumers more options in terms of how they interact with a virtual assistant, it means that we can go beyond augmented reality to showcase products in the most relevant contexts. For example, rather than simply overlaying new clothing onto an image of a customer at home (augmented reality), a virtual stylist, supercharged with generative AI, might show the customer wearing a new summer outfit on the beach in Barbados, outside the hotel they’ve booked for this year’s holiday, really helping them visualise before buying. Data makes the difference While existing LLMs are incredibly powerful and have many uses on their own, they don’t have access to personal data and are limited in how far they can provide a bespoke experience. By training a model on proprietary data and incorporating customer data (with permission, of course), retailers can take the benefits of these tools a step further, providing a personal experience second to none. A fashion retailer, for instance, could create a personal stylist who knows what clothes a regular customer looks good in, based on a combination of purchase history and back-and-forth interaction with them. The more frequently the customer interacts with the virtual assistant, the more accurate and valuable its output will become – encouraging customer loyalty in the process. In fact, optimal data usage is make or break for the success of a virtual assistant project for any retailer. Only by understanding exactly what information is available across the organisation as a whole, including where it is stored and how it can be accessed, can they hope to extract maximum value for it and, in turn, pass this value on to the customer. This, then, is the future of shopping. It sees a return of the back-and-forth interaction of the past, which was lost in the throes of the online revolution. Just as vitally, it’s a future where high quality experiences - once the preserve of the rich - are made available to all. If the future is about accessibility, then AI is the key to unlocking it. And those retailers who adopt this tech early, who understand how to leverage the vast amounts of data currently at their fingertips, are the ones who’ll succeed. Sam Dods, Head of Mobile Engineering at Kin + Carta