Article
March 16, 2026 · 4 min read timeShopping has always been an active journey. Consumers make decisions based on search and comparison results when shopping online and in-store. With the rise of AI agents, these defaults are starting to change. What will the future of retail look like? At SOK, Mikko Hänninen VP, Development & Data is responsible for development and data in areas including assortment planning, pricing and space management. Ville Henriksson is a Senior Service Designer at Nitor.
A fridge that automatically orders refills. A price‑hunting agent scanning the web for the best sneaker deals. A personal assistant that orders winter clothing just in time for a ski trip.
The technology behind agent‑driven purchasing is often referred to as agentic commerce. With this model, not only does the purchasing process become faster, but it is delegated to an AI agent that operates independently within predefined boundaries.
When an agent can complete the entire purchase journey and pay for products on its own, we’re no longer talking about an intelligent search tool. It is commerce carried out on behalf of the consumer. This simultaneously changes the logic of buying, brand loyalty and adds new aspect to the design work.
Daily life where shopping fades into the background
Agentic purchasing will first become visible in everyday routines. Shopping for consumables, repeat purchases and other clearly defined purchase decisions will shift to agents almost unnoticed. The shopping basket is replenished before products run out, while the agent negotiates on the consumer’s behalf to assemble the best possible basket.
When the agent has access to payment details, delivery preferences and return policies, purchasing takes a step away from direct human control. In terms of certain categories an agent is simply a more efficient and objective decision-maker than a human being.
However, not all shopping is the same. Consumers want to automate some purchases completely, while retaining more control over others. For more expensive and emotionally charged purchasing decisions the agent initially acts as an assistant while providing most of the groundwork – searching, comparing and making recommendations.
Essentially, the agent optimises the whole basket: price, availability, delivery and reliability.
While automating everyday grocery shopping is ultimately straightforward, buying a new sofa is different. In situations like these, consumers likely want more decision‑making power, as vague personal preferences are harder to articulate to an agent.
Agents read structured data
When purchasing is no longer guided by keywords but by consumer needs, discoverability changes fundamentally. Search engine optimisation (SEO) based on individual search terms starts to lose relevance.
An agent does not search for an “affordable suitcase". Instead, it works towards a defined intention and use case. It relies on data to find a lightweight, durable suitcase with fast delivery from a reliable vendor. The user’s existing preferences also play into the results.
Agent‑optimised visibility is based on a much broader set of parameters than traditional search visibility. Factors such as return rates, damage history, delivery reliability and customer reviews carry significant weight alongside product data.
In this world, data is not a support function for marketing, but the foundation of all visibility. An agent does not browse an online store; it reads structured data. Taxonomies, attributes and structured product information determine whether a company’s products are even considered by an agent. From a service design perspective, optimising the data and APIs takes precedence over optimising the storefront. This ensures that the agent concludes that the retailer is a credible and trustworthy option.
Structured data requires explicit definitions. This may feel clumsy at first, but agents cannot make reliable decisions without them. For a human, it is obvious that a banana is a vegan product. The same isn’t necessarily true for an agent.
The data‑driven decision‑making by agents can also forge a path through marketing noise. If marketing messages conflict with actual data, the agent reacts immediately. Trust is not built on promises but on consistency. If the data contains discrepancies or is fragmented, the store loses credibility in the agent’s eyes.
Loyalty changes, but does not disappear
As human involvement in purchasing diminishes, what happens to customer loyalty? In a worst-case scenario a company becomes little more than a warehouse where agents do shopping based purely on price.
Brand loyalty likely remains strong in categories such as groceries. An agent will not arbitrarily switch products if the consumer’s preference is clear. However, loyalty to a specific retailer weakens if the agent finds the same offering with better execution elsewhere.
Traditional memberships and loyalty programmes also lose impact if an agent can automatically log in, claim the benefits and move on. Communication designed for humans no longer works in the same way when the agent is the one doing the shopping.
In the future, loyalty is created through the overall experience: availability, services built around the product and the functionality of the entire purchase journey. A well‑designed ecosystem remains competitive.
The future is here soon
Agentic commerce is not a distant scenario. Language models are already widely used to support shopping, and the next step is authorising them to act on behalf of the consumer.
From a business perspective, the message is clear: if preparation starts now, you are already running late to the game. Consistent operational product data and an agent‑friendly architecture are basic requirements, not competitive advantages in the agentic AI era.
Everything starts with data. The companies that land on top will undoubtedly be the ones that have access to customer data enabling personalisation and recommendations, a rich product assortment and a service ecosystem that caters to excellent customer experiences.
Would you like to upgrade your commerce to meet the new requirements? Get in touch!
Written by
Mikko Hänninen is Vice President, Development and Data at SOK leading data, development, and processes across S Group's grocery and non-food category management and procurement (sourcing, assortment, pricing and macro/micro space management). Prior to joining SOK Mikko worked with a number of Nordic retailers and suppliers to shape digital retail strategies and accelerate digital growth both as an academic and consultant. Mikko holds a doctoral degree from the Aalto University School of Business. His research has been published in leading strategic management journals such as Journal of Management and Journal of Management Studies.
Ville Henriksson is a multidisciplinary service designer who specialises in retail solutions. In his work, he operates at the intersection of user experience, business requirements, and product data - particularly how intelligent systems enable new forms of commerce experiences. In his free time, he explores the world of mid-century Finnish design and dreams of being a furniture maker in his workshop.
