Coming into this assignment I had preconceptions of AI in business, having worked myself in a corporate setting and seeing the impact of AI in our jobs within even the past year. Yet with this in mind, I treated the assignment at face value.
The pain point we wanted to address came from my experience shopping at Albert Heijn, where I find myself roaming around looking for meal ideas, or constantly returning through the store once they arise. Thus the idea formed: let’s help shoppers have a better experience in store. Our solution focused on an AI ecosystem between the app and the handscanners to help plan your meals and shopping list, and then in the store, to be effectively guided through the aisles and receive customized assistance. The main solution showcase focused on the in-store navigation system and the personalized assistance within the handscanner.
My contribution focused on the solution. On one hand I designed the solution and everything it would involve from a conceptual perspective, while also handling the prototype: the AH Smart Shopper. The process itself proved incredibly valuable, as visualizing the solution, not only helps in iteration, but also highlights the customer experience and missing elements, much better than just discussing it at a conceptual level. A significant contribution to the final solution can be handed to the prototyping process as it pin-pointed key elements that had to be improved or included.
One such feature was the “This product is missing” feature. A customer may run into a product being out of stock without this being reflected within the store’s inventory system. Thus, there must exist an option to receive a recommendation for a product that closely matches the missing option. In a similar vein, we also wanted the solution to manage in real time the store flow and prevent over-crowding. Apart from the routing being done such that customers do not focus in the same area simultaneously, using personalized offers and suggestion can serve the same purpose if the customer accepts them. The system thus tailores these recommendations not only on the customer history and shopping patterns, but also in the context of store flow and crowded “hotspots”. This second way of preventing crowdedness came as a result of protopying, as it had not been in the initial concept.
The survey we ran was integral to understanding whether the way we selected the features of the ecosystem was considered valuable by AH customers. An important element we found, was that the way we designed it to suggest alternatives rather than automatically replace the missing product was the preferred behavior. The general findings of our survey were that the customers found the recommendations and navigation useful, yet it was important that they have full autonomy over their shopping. Thus it was important to ensure that recommendations were spaced out and relevant, such that the system felt like an assistant rather than a marketing tool. We also tested whether users would become over reliant in cases of allergies, which was not the case, as an overwhelming majority did not accept a recommendation violating an allergy. It thus confirmed to us that a hard allergy filter is required, but it also shows that the customers have the necessary agency and autonomy, further reinforcing against automatic replacement in the shopping list.
Our team approached the assignment with a collaborative independence approach, where every member had a set of responsibilities to focus on, with a clear goal and being focused on the solution. This allowed us to work independently while ensuring that we were keeping a closed and quick feedback loop. Specializing our roles helped structure the assignment in a similar fashion to the way a company approaches such a project, allowing us to play on our strengths and prevent possible frictions of communication. Owning our segments allowed us to focus on our work and align when necessary, creating an efficient team effort. Finally, keeping feedback open and establishing that our goal was to have the best possible solution both for customers and Albert Heijn, enabled us to manage and discuss diverging opinions effectively and objectively.
I have never been an AI skeptic, but I have always believed it is over stated in its current form. This is the perspective I came in with for the course and the AI Strategy Lab. Seeing the rise of AI within my current company, its impact, benefits and shortcomings has formed this perspective. Within this assignment, there are 2 lanes which had an influence on my perspective: using AI tools for prototyping and the role of AI within large enterprises such as Albert Heijn.
On the prototyping side, Lovable was used for creating the prototype. It is very impressive to have access to a tool which can have a working prototype in a matter of hours and empower the average user to create such solutions. Yet what became evident is that even though such solutions are very capable when it comes to creating the prototype itself, they struggle with the fine details. What was very surprising in the process was the effort required to iron out the last details. The final part of adding small changes or fixing minute elements which were not properly represented was the most time intensive and it accounted for the largest consumption of tool credits. This experience contextualized the idea that AI should not replace employees but be a tool used for their work. Many business owners consider they can use AI instead of hiring a professional, which may be true, but the quality of the end product and time investment required, could likely indicate that a professional empowered by AI is a much better choice.
Having finished the assignment, I find it that AI should be treated as a tool and have implementations which are reliable and designed to be resilient. Many companies rush into their AI implementations which can leave them exposed in the event of a market downturn, especially as AI remains an expensive solution. Ensuring that AI is a driver of efficiency and quality, rather than a patch on a larger issue or a reason to make operations leaner, should be the approach taken. Driving AI adoption for the sake of adoption without real utility will likely prove to be unsustainable, which is something that is becoming more and more evident. Taking everything into account, my current view is that when AI is treated as a solution to what it can actually solve, rather than a magic fix that applies to everything, it creates value in a resilient manner and brings forth real benefits for everyone involved. As a result, our proposed solution tried to focus on using the strengths of AI, while allowing for customer autonomy and ensuring that overburdening of users is prevented.
