AI in retail: know your craft, then automate | D-Congress 2026
Anastasiia Starchenko
At D-Congress 2026 in Stockholm, Ingrid's CPTO and Co-founder Piotr Zaleski took the stage twice — first on March 4 and again on March 5 — and both times he returned to the same uncomfortable truth: the biggest obstacle to AI in retail isn't technology. It's us.
Piotr was joined on stage by Jacob Rastad (CTO, Nordic Nest), Tim Richardson (CEO, ROIROI), Hanna Moisander (Chair, Lager 157), and Fredrik Hjelm (CEO, Voi).
The broken foundation problem
Across two panel discussions spanning topics from AI-native agents to the future of retail automation, the panels painted a picture that many in the audience likely recognized. Companies have spent years digitizing processes that were broken to begin with. When they then try to layer AI and automations on top, the results disappoint — hallucinations, errors, lost trust.
"It's not AI that's the problem, it's a human problem. We've been a bit lazy on digitalization, taking shortcuts and going for quick wins instead of safe wins."
As a result, organizations are now losing faith not in the innovation itself, but in their own ability to wield it. Trust erodes, not because the tools don't work, but because the groundwork for retail automation was never properly laid.
Know what not to automate
The prescription was refreshingly analog for a tech conference: start with pen and paper. Understand what your company actually does, step by step. Identify what constitutes real craftsmanship — the things that make your brand impossible to replicate — and separate that from everything that could and should be automated.
"We want to know the craftsmanship. We want to know what to do and how to do it. Then we can decide how to digitalize and automate."
The unsexy work — getting box sizes right, placing the right inventory in the right store, optimizing what the customer never sees — that's where AI creates real, lasting value. Not in the flashy demos, but in the seamless operations behind them.
The Feedback-Action Framework
The talks also zoomed in on what separates effective AI implementations from hype. It comes down to two things: a feedback loop and an action loop. The best vendors in a retailer's tech stack are collecting data and acting on it.
"You need to look at which companies in your tech stack already have the feedback loop and the action loop that you can utilize.”
They span entire domains, bridging what happens after a purchase to what's promised before it. These vendors are the ones that will define the next era of retail technology. The post-purchase experience — delivery tracking, returns, fulfillment accuracy — is where most customers form lasting opinions about a brand. Yet it's precisely this domain that gets overlooked when companies rush to automate.
Automate the boring, protect the brilliant
Following both talks, the overarching argument was clear. The companies that will win AI adoption are the ones who first understand what makes them special, then ruthlessly automate everything else.
Back-office operations, business admin, repetitive workflows — no company does these because they want to. They do them because they have to. With AI-powered tooling now capable of delivering customized internal software at a fraction of traditional costs, the excuses are running thin.
The future belongs to retailers who protect their craft and delegate the rest to machines smart enough to handle it. First though, pick up a pen, understand what you actually do, and only then decide what the AI should take over.
Learn more about Ingrid AI →
Ingrid AI orchestrates delivery windows, pricing and carrier optimization based on real delivery performance, instead of static rules. Each completed order makes the next delivery decision smarter. Intelligence that moves delivery from fixed configuration to adaptive execution across the entire delivery journey.
FAQs
When does the 14-day withdrawal clock start?
For goods, the 14-day period begins the day after the shopper takes physical possession of the order, which in practice means the day after delivery. If multiple items in one order are delivered separately, the clock starts the day after delivery of the last item.
Is the withdrawal the same as returns?
No. The right of withdrawal is a statutory right under EU law; a return is a commercial process. Labels like ‘Return’ or ‘Get a refund’ do not qualify on their own — the withdrawal option has to be clearly and unambiguously labeled as such. Some teams are leaning toward separate flows, others toward one combined flow. Both can work, but only if withdrawal remains a clearly labeled, unobstructed step within it.
Do shoppers have to give a reason to withdraw?
No. Providing a reason is not required by law. Reason fields in the withdrawal flow must be optional and must not delay or block completion of the withdrawal. Designing the interface in a way that pressures shoppers to provide a reason is treated as a manipulative pattern under the directive.
Does this apply to retailers based outside the EU?
Yes — if they sell to EU shoppers. The directive applies to distance contracts with EU shoppers, regardless of where the retailer is established.
Related blog posts
EU Right of Withdrawal: Your implementation guide for 2026
How to choose the implementation path that fits your stack and the pitfalls to watch out for
EU Right of Withdrawal: What retailers need to know in 2026
Here's what the rule says, what counts as compliant, and where most teams are stuck.
How OSPREY LONDON turned delivery into a profit driver
A conversation with Ben Jones, Head of E-commerce and Technology at OSPREY LONDON
