For decades, IKEA has persuaded millions to buy furniture that comes in parts. Its PAX wardrobe, designed to be built with little more than an Allen key and some faith in pictograms, shows the appeal. It is cheaper than premade alternatives and endlessly reconfigurable, so long as the builder keeps the particleboard frame stable. That formula has helped turn the Swedish retailer into the world’s largest furniture business, with sales of €45.1bn ($48.5bn) last year.
Now IKEA is applying the model of standard parts and local assembly to artificial intelligence (AI). Just as its customers put together a PAX that suits them, IKEA’s country teams and shop-floor “coworkers”, in its own jargon, are encouraged to slot AI into their daily work as it best fits—within clear guardrails. With some studies suggesting that up to seven in ten of corporate change programmes fail, often due to employee resistance, IKEA’s approach to AI shows that success lies less in acquiring a fancy model than in creating the conditions for people to assemble it into something useful. Its method rests on three pillars: defining intent with clear design principles, building universal capability and structuring the organisation for grassroots innovation.
Set purpose and boundaries, then accelerate
There are few areas of a modern firm where AI cannot be applied to some degree. But doing a little of everything is a good way to achieve not very much. This is why IKEA has a process for deciding where to deploy AI. “What’s the point of using AI and how does it really help the core mission and vision of what IKEA wants to get done?” asks Parag Parekh, the firm’s chief digital officer, in conversation with Economist Enterprise, parts of which you can listen to across this article.

Part of IKEA’s answer is to rule out AI uses that would jar with its core values. “We’ll never use AI for tracking our coworkers or our customers,” Mr Parekh says, nor will IKEA adopt AI hiring tools until it is confident they do not introduce any bias. Drawing such explicit boundaries around the tech helps large organisations stop scattered experiments from going awry. But it also creates a safe space for employees to accelerate AI use and innovation—and IKEA’s coworkers are encouraged to do just that.
But even with boundaries in place, priorities are still needed. To focus on the right AI projects, IKEA has repurposed its time-tested product-development philosophy of “democratic design”—the balance of function, form, quality, sustainability and low price. Applying it to AI development ensures that AI pilots are technically feasible, as well as aligned with IKEA’s business goals and its distinctive corporate culture.
Consider democratic design as applied to IKEA’s AI use in the supply chain. Mr Parekh starts by recognising that “in some respects, IKEA is more a supply-chain organisation than a retail one”. Pilot tools successfully sharpened demand forecasts and routed inventory to the right depots, demonstrating the function of using AI. They were also specifically designed for warehouse managers and store teams—a form that fits daily routines. As the algorithms proved robust, operations gained in quality, resulting in fewer stockouts, steadier distribution and faster inventory turnover. And because the optimisations AI brought also cut redundant lorry trips, the system advanced sustainability while trimming prices.
Other firms might take note of IKEA’s process. Research from S&P Global, a consultancy, finds that over four in ten firms worldwide scrapped most of their AI initiatives in 2025, up sharply from fewer than two in ten the year before. Too often executives set lofty aspirations, pursue a flurry of AI pilots and hope something useful emerges. The result is usually investment with disappointing impact. IKEA’s approach can be useful: by anchoring AI in purpose, setting clear boundaries on its use and concentrating efforts where value and values align, the firm manages to build useful AI solutions.
Make the training universal, not the playbook
Technology is likely to stick if the people who use it feel part of its success. IKEA has therefore made AI literacy “like health-and-safety training”, says Mr Parekh—available to everyone and compulsory for newjoiners. AI training programmes are being rolled out in over 31 countries, with the aim of reaching 70,000 staff and leaders by 2026. “Our goal is to make sure that we literally do not leave any coworker behind,” says Mr Parekh.
Levelling capability across workers matters because AI will not lift every worker equally. In many firms, it will help novices most by capturing the tacit know-how of experts and making it available to everyone (read more about who AI helps most in this series).
Some parts of AI training will be similar across firms. How to use copilots to draft, analyse and summarise is fast becoming basic literacy. So too are habits of verification—essential when using generative AI systems prone to making things up—and strategies for dividing labour between human and machine (read more about “centaur” teams in this series).
Yet the bulk of a firm’s AI training must be tailored to what its employees do, as well as its culture. In a warehouse, IKEA staff learn how AI guides logistics; in a design studio, how it could help with creative tasks. What binds the programme is what Mr Parekh calls doing it “in an IKEA way”, embedding the firm’s culture into every training module. Generic instruction that does not vary by role, market and business need risks making AI tools feel irrelevant.

Even then, some pushback is inevitable. “Resistance,” Mr Parekh suggests, “is not directed at the change itself, but at what you are losing due to that change.” Fear of diminished control, status or routine is rational; treating it as such shifts the corporate response from imposition to support. IKEA’s method has been to demonstrate, in practice, that AI is meant to augment, not replace (read more on augmentation with AI in this series).
IKEA’s planning services illustrate the point. A designer who once spent five to seven hours creating a 3D model of a customer’s room completes the task in under an hour with the help of AI. Freed from drudgery, the planner can serve more clients and devote time to the creative exchanges customers value. Productivity rises, jobs remain, and the service is, as IKEA would have it, “democratised”.
The outcome at IKEA is a workforce steadily levelled up, not dazzled, by new AI tools. Literacy is spread thin but wide, training is iterative and resistance is mined for insight rather than treated as obstruction. For other firms, the lesson is plain: the culture of learning around AI may prove just as decisive as the algorithms themselves.
Share kits and assemble locally
If the first task is deciding where AI belongs, and the second is preparing people to use it, the third is organising how ideas travel. IKEA’s model is a light centre with heavy edges. “You have a central team in the global organisation, which sits partly with Digital, but partly also in People & Culture,” says Mr Parekh, suggesting that it is a design choice that underscores the initiative’s socio-technical nature from the outset.

The centre equips and enables, creating and distributing standardised “shared kits”—the AI equivalent of flat-pack components, including templates and presentations —so that each of IKEA’s 30-odd markets starts from the same high-quality base. It also tracks adoption, using telemetry to see how tools are used—around 170,000 coworkers have access to the internal copilot, with over 26,000 using it monthly—and relies on a network of “ambassadors in every country” to collect feedback and share lessons. With this data, the central team can see what works, spread it fast and steer experiments back on course when they drift. But it is at the edges where a lot of the AI innovation happens. Country teams act as “receivers”, taking the shared kit and adapting it to local needs, allowing new uses to emerge from the bottom up.
The ultimate proof of this model’s success comes from the tools that have emerged from country teams. Mr Parekh emphasises how coworkers have built assistants from IKEA’s own data: one draws on eight years of “dense Life at Home reports,” turning them into a simple, queryable resource for all staff. Another, a “furniture dealer” assistant, channels the wisdom of founder Ingvar Kamprad’s handbook to answer questions about corporate culture and values.
Building AI-ready firms
Flat-packs helped make IKEA the world’s largest furniture-maker, and the firm is now using the same approach to AI, showing how even the smartest algorithms work better when everyone has a hand in tightening the bolts. Building AI-ready firms, the first instalment of the Enterprise AI in action series, also includes an article describing what Erik Brynjolfsson, a leading economist of the digital age, thinks it will take for a firm to gain from the technology, and an article on how LVMH, the world’s largest luxury group, is using AI to strengthen its distinctive character.
Sources
McKinsey & Company. Changing change management. July 2015. Available at: https://www.mckinsey.com/featured-insights/leadership/changing-change-management
S&P Global Market Intelligence. AI experiences rapid adoption, but with mixed outcomes – Highlights from VotE: AI & Machine Learning. May 30 2025. Available at: https://www.spglobal.com/market-intelligence/en/news-insights/research/ai-experiences-rapid-adoption-but-with-mixed-outcomes-highlights-from-vote-ai-machine-learning










