Jumbo has launched a pilot project using agentic AI. The Dutch supermarket operator aims to support online customers with their shopping using its own AI agent. The tool is designed to inspire shoppers and help them put together their shopping baskets, according to Dutch trade magazine Distrifood.
Kevin Rasenberg, Head of E-Commerce, and Niels Nijdam, Head of Store Technology, Data and AI, are responsible for the project at Jumbo. Their task is to identify areas of application for agent-based AI across the entire company in the long term. As a first step, the team has begun testing the use of an AI agent for online shopping. This is intended to help customers put together their shopping baskets.
Jumbo is currently testing the ‘Agentic Shopping Assistant’ within its ‘Friends and Family’ panel, which comprises just a few dozen people. The Jumbo online test panel, with around 10,000 members, will also be able to use this feature shortly. “This pilot project is currently still very much focused on providing inspiration and the first step in shopping support,” explains Kevin Rasenberg in an interview with Distrifood. Should the pilot project prove successful, the scope may be expanded.
Customers are getting used to conversational shopping
Jumbo’s test customers can use a chat function to discuss their shopping with the AI agent. For example, they can ask for inspiration for a specific occasion or for recipes. In doing so, they can specify their preferences or dietary restrictions and request, for instance, only organic, vegetarian or gluten-free products. They can then add the suggested products to their basket with a single click.
“Customers are increasingly getting used to conversational shopping on AI platforms,” explains Kevin Rasenberg. This shift in customer behaviour has made it clear to Jumbo that the company needed to act quickly. The number of visits by agents to their platform has slowly begun to rise. That is why Jumbo is increasingly focusing on agent-based AI.
An opportunity to learn
Initially, Jumbo is working on its own chatbot for its website. This must go through the same process as an external chatbot, explains Niels Nijdam: “We must first open our platform to our own chatbot; after that, we can easily open it up to external ones. This step helps us lay the groundwork for that.”
Jumbo also sees the pilot project as an opportunity to learn how agentic AI can support internal processes. Kevin Rasenberg cites, for example, staff in stores and distribution centres: “How can we provide them with the answers they need more quickly?” Niels Nijdam points, amongst other things, to the efficiency of supplying goods to the stores. New staff might soon be able to use an AI agent for their queries, rather than having to work their way through a comprehensive operations manual.
Avoiding dependence on individual technology providers
The project managers did not disclose possible technology partners for the project, however, emphasise that they deliberately selected the most suitable platform for each use case. They set it up in such a way as to remain flexible and avoid becoming dependent on a single provider. Depending on the problem at hand, we chose whatever worked best for the customer at that time. Behind this lies a broader objective: the development of technical expertise within Jumbo. “We believe in building a technical department that truly understands what the customer wants,” says Niels Nijdam.
