We AI Like Toyota. Not Like Klarna.
Klarna replaced 700 customer service employees with AI, declared it a win, and then reversed course eighteen months later and started rehiring. The CEO’s own explanation was that cost had been too dominant an evaluation factor — a careful way of saying they went too big, too fast, before they understood where it breaks.
I’m an AI advocate and have been pushing it into our operations for years. But the mistake Klarna made isn’t moving too fast. It’s moving too big before you understand the edges.
When you scale AI deployment before you’ve stress-tested it, you don’t find the failure modes until you’ve already built your operating model around them. The metrics look clean. The output looks right. What you don’t see yet is where the confidence is false, where edge cases fall apart, where a human should have stayed in the loop.
We have a deeply ingrained Kaizen culture at iE — continuous improvement in small increments, systematically applied — and it turns out to be a near-perfect framework for AI adoption. We identify a specific workflow component, understand it well enough to know where AI can credibly own it, deploy it, find the edges, then move to the next one. It’s slower than going in with a hacksaw. That’s also why it works.
The other pattern I’ve noticed is that companies moving too fast tend to go for the big visible things first — headcount reduction, cost centers, customer-facing automation. The board deck stuff. They skip the smaller workflow applications because those aren’t a headline, and in doing so they skip the part where you actually learn what AI can and can’t do. When you skip that, you’re making large bets on a system you don’t fully understand yet.
The framing we operate from is that AI should empower the people we have to do more, not replace them. We’re a growth company — AI is fuel for that, not a cost-cutting lever. Every efficiency we find creates capacity for the next thing we want to build, the next market we want to go after. If you’re growing, that’s the right way to think about it. The companies that got most excited about AI-driven headcount reduction were mostly looking for a short-term number, and Klarna is a pretty clean case study in where that lands.
We’re behind on some things — I’ll say that directly. But the approach we’re taking means that when we make a bigger bet on AI, we’ve already found the edges at smaller scale. That’s the version of this that actually compounds.