IKEA's chief digital officer, Johanna Roos, offered up a single sentence in a McKinsey interview earlier this year: "The risk of trying to do everything and ending up doing nothing." It's a surprising line coming from the person in charge of digital transformation at a company that runs more than 400 stores across 60 countries. At that scale, you'd assume there's a dedicated AI team and a budget for outside consultants. And yet the first thing she reached for was a question: What do we do first?
Agentic AI is the technology direction global companies are experimenting with most aggressively right now. It carries out multi-step tasks on its own, connects to outside systems, and makes judgment calls as it executes. If a chatbot operates at the level of answering "What's the return policy on this product?", agentic AI takes that same question, pulls up the customer's purchase history, checks the logistics schedule, and processes the return all the way through. IKEA is on a journey to apply this technology across the entire customer experience. And the first wall the company ran into along the way wasn't a technical one.
Even 400 stores couldn't dodge the priority problem
The heaviest part of what IKEA's CDO shared in that McKinsey interview wasn't about agentic AI as a technology—it was her answer to how they decided what to adopt first. The moment you introduce agentic AI, a flood of newly possible things appears all at once. Automated customer service, inventory forecasting, delivery management, personalized product recommendations, employee training support. Even at IKEA's scale, you can't push all of it forward simultaneously. The CDO concentrated the company's resources on building the criteria for what to choose first.
That choice runs on more than technical feasibility—organizational culture, the state of the data, and the trust relationship with customers all factor in. Agentic AI reaches into data, calls external systems, and sometimes makes decisions that handle customer information. A single wrong move in that process hits brand trust directly. IKEA said its first consideration is the quality of the customer experience and the preservation of trust. Which AI features get switched on is decided beneath that standard.
Once agentic AI is live, you end up with a complex system in which each agent connects to other agents. One agent checks inventory, another coordinates the delivery schedule, and yet another sends a message to the customer—a chain reaction. For that chain to work properly, the data quality and system stability at every connection point have to be in place first. That technical reality is the backdrop to the CDO's warning about "the risk of trying to do everything and ending up doing nothing."
The gap in scale doesn't block the lesson
It's a fair objection that you can't put IKEA-level infrastructure on the same plane as the reality of a Korean solo operator or a one-person PM. IKEA has a team of data engineers, a CDO, and a budget for outside consultants. To someone running a business alone, "build an agentic AI strategy" can sound hollow. Dig deeper, and IKEA is a company that has accumulated data for decades. When its agentic AI queries inventory, underneath it sit standardized product codes and a global logistics database. For someone working solo, building that foundation in the first place can be a far longer and harder task. The simple comparison—"IKEA did it, so we can too"—creates an optical illusion that makes the barrier to execution look lower than it is.
And yet the priority confusion IKEA experienced isn't proportional to scale. The smaller you are, the less room you have for mistakes, which means a misjudged priority turns fatal even faster. If you're a solo PM who tried automating customer service first with Claude or GPT, and months went by without your actual working hours dropping, then the IKEA CDO's warning won't feel unfamiliar. The more AI tools pile up, the harder it gets to know where to begin—and that feeling shows up regardless of how big the organization is.
What a solo worker should check first before using agentic AI
So where can you actually start?
The real value of agentic AI lies in delegating repetitive, multi-step work. For someone working alone, the biggest time sinks usually fall into three areas: gathering information, drafting, and organizing internal communication. Building an agentic AI flow in one of these three areas first is the practical equivalent of the "what to choose first" decision the IKEA CDO described.
If part of your job is compiling industry news each week and sending it to a client, you could try wiring that entire flow over to agentic AI. The AI collects information on specific keywords, summarizes it, and drafts it in the right format—leaving you only to review and send. Expand to the next area of work before this one is fully stabilized, and you land in exactly the situation the CDO warned about. Run several things at once while one isn't working properly, and you can't even tell where the error came from.
The better AI tools get, the more the skill an individual needs to develop isn't the skill of operating the tools. It's the ability to design which tasks to hand to AI and which judgments to keep for yourself. The fact that a CDO leading an organization of hundreds of AI specialists personally set the priority criteria points in this direction. The more complex the chain of work AI performs, the more the judgment of the person who first designs that chain determines the quality of the output. I'd argue this is the most practical capability an individual needs to develop in the age of agentic AI.
IKEA's agentic AI journey looks like the dazzling transformation story of a global enterprise, but the question at its center applies just as fully to a small operation. What do you do first, and what do you push to later? Get that judgment wrong, and even good tools just accumulate without direction. The more tasks you can hand off to AI, the more the work left for a human narrows down to one thing: deciding which of those tasks to assign first.
