Last year, Klarna bragged about their AI customer service tool saving them from 700 human hires and cutting response times by 82%. Now founder Sebastian Siemiatkowski has had an "epiphany," declaring that in an AI world, "nothing will be as valuable as humans."
Has the pioneer of AI operationalization - the company that ditched Salesforce and Workday for DIY AI and celebrated shrinking their workforce by 1,200 people because of AI - had a genuine change of heart? Or have they uncovered something far more interesting?
There is a fundamental question at the heart of this: Does AI lift the floor or raise the roof in organizations? Is it the great equalizer that democratizes excellence, or is it the great divider that turns the gap between the best and the rest into a chasm?
Early evidence pointed to AI as the great equalizer. Customer service was the perfect poster child: MIT researchers found AI boosted new rep productivity by 34% while barely moving the needle for veterans. The technology was bridging the experience gap, just as promised. But that was only part of the story.
Perfect Isn’t Always The Point
Shift to more complex domains, and a different picture emerges. When MIT gave AI tools to materials researchers, it doubled the productivity of top performers while leaving the bottom third unchanged - The Economist found this pattern repeating across many sophisticated use cases. AI wasn't closing gaps - it was widening them.
Why? Because AI isn't infallible - it's a powerful but imperfect tool that requires skilled hands to yield results. Take Deep Research, which recently caught Benedict Evans's attention. In analyzing an OpenAI showcase report, he found glaring errors: iOS's Japanese market share was reported as 69% when the source showed 36% (the real number is closer to 47%). To Evans, an analyst whose reputation depends on accuracy, such errors make the tool unusable.
But, perfect isn't always the point. There are countless scenarios where getting to 85% accuracy quickly, with human judgment filtering the output, creates massive value. This is what Klarna's CEO has finally grasped, and it has profound implications.
Imperfect AI: Your Perfect Advantage
In the last year, Klarna increased revenue and profit while their headcount dropped from 5,000 to 3,800 through attrition. Cursor hit $100M in ARR with just 20 employees. AI is enabling smaller, more capable teams to drive unprecedented results - but only when those teams excel at wielding it.
This convergence - of AI amplifying skill gaps while enabling leaner operations - points to an inescapable conclusion: success in the AI era hinges on attracting and retaining high-caliber individuals who possess the judgment, taste, and intuition to harness AI effectively. The challenge is that these traits are far harder to assess in an interview than years of experience or technical skills.
Companies must now fundamentally rethink how they identify and retain talent. The traditional playbook of screening for experience and technical skill still matters - domain expertise helps tame AI's wilder tendencies - but it's no longer enough. We need new frameworks to evaluate judgment, taste, and intuition, qualities that increasingly separate exceptional performance from mere competence in an AI world.
The good news is that we are still early in this phase shift. Organizations that focus now on building these talent identification and retention capabilities will create lasting advantages. The race isn't just about implementing AI - it's about building teams that can turn AI from a commodity tool into a genuine competitive weapon.
For those looking to build world-class organizations, the mandate is quite clear: get extraordinarily good at finding, attracting, and retaining high-caliber individuals who can harness AI's potential. The technology might be available to everyone, but the talent that maximizes its impact? That's anything but a commodity.
