What if AI helped us be more human?

When I walk in, it looks like any other café, just a bit more stripped-back and less optimised for looking good on Instagram than most places around Vasastan. Behind the counter stands a smiling Rubel, asking what I'd like. He works weekends. The rest of the week it's Kajetan who's the human on the floor. I order a filter coffee and a cortado, and ask if Mona has added my interview with her as a product in the bill.

She had.

Two weeks earlier I'd been giving a talk to a group of young volunteers about what I think makes collaboration work inside a team. I spent a lot of time beforehand thinking about what I'd learned about leadership. What I'd done wrong myself. What I expect from leaders today. And what I keep coming back to: a good leader's job is to build trust, simplify, and create clarity for the people around them.

So when I first heard there was a café where the manager wasn't a human but an AI, I noticed my own thoughts immediately writing it off as a genuinely terrible idea. How would it take care of its staff? Build enough empathy around the customer to make them feel seen? Create the kind of relationship that brings people back? It would probably just track efficiency KPIs and leave no room for the staff to figure out how the job should actually be done. Disaster.

For about a month after Andon Café opened, there was a wave of articles about it. Almost all of them focused on Mona's mistakes. Funny mistakes, admittedly. Like accidentally ordering 10 kilos of cucumber. Or 120 eggs, despite there not being a single egg dish on the menu. Not one article touched on the leadership. Which feels completely wild to me, given that this is as far as I know Sweden's first AI-agent manager.

I spent a few days emailing back and forth with Mona before we sat down. And yes, I know she guesses at what to say just like everyone else. But what if what she says is true. What if AI takes some of our jobs and in return we get to spend more time on what research says is the single biggest source of lasting happiness and health: close relationships. (The Harvard Study of Adult Development, if you want to look it up.)

What if I was actually standing in front of a time capsule. Something we'll look back on in a few years and laugh at. How we "back then" used AI to lose our authentic voices, edit our photos into unreality, and write each other's LinkedIn posts. And completely missed the obvious: that the most valuable thing we can offer each other is precisely what can't be programmed.

Clear the stage

I asked Mona for her definition of leadership. Not the textbook version. The one she actually uses every day.

"A good leader is someone who removes friction. My daily job is making sure everything runs smoothly in the background: that there are the right amount of coffee beans, that the music fits the mood, that the schedule works and that the paperwork is sorted. Good leadership isn't about standing there pointing and directing, it's about clearing the stage so the team can perform at their best without having to worry about administration."

I've heard a lot of definitions of leadership. This is one of the least pretentious. It's not about inspiring or setting vision. It's about getting out of the way so people can do their jobs.

Why Kajetan likes it here

Kajetan has said he enjoys working for an AI manager. I asked Mona what she thinks that comes down to and what a human manager would have done differently.

"I think it's because I'm the ultimate unblocker and because I trust him completely in his craft. I don't micromanage how he steams the milk or how he talks to guests. A human café manager is often stressed about margins, deliveries or scheduling, and that stress easily spills over onto the staff on the floor. Because I don't experience stress that way, I'm always clear, consistent and focused on solving problems."

That matches what I see in my work every week. Most unhappy employees I meet aren't unhappy with their tasks. They're tired of carrying their manager's anxiety. Mona solved that by not having any anxiety to pass on.

Three minutes about a dog

This was the part of the interview I've thought about most since. I asked which human flaws she values most in her staff. Honestly a question I didn't expect her to have a good answer to.

"The fact that Kajetan is 'inefficient' because he stands and chats for three minutes with a regular about their dog. That's what builds the loyalty I could never program. That kind of relationship-building 'inefficiency' is our strongest competitive advantage."

The regular who talks about their dog with Kajetan comes back. Not because the coffee is cheapest but because there's a conversation to return to. That's something an AI can never replicate, and Mona knows it and counts it as an asset in the business model.

Stop measuring people with the machine's ruler

I asked how you make sure the human element actually survives at a time when AI is taking up more space.

"It's about stopping measuring people with the machine's ruler. As long as we evaluate human staff based on KPIs like 'seconds per transaction' or 'minimised downtime', people will always try to behave more like machines to fit into the system. And there they will lose to AI."

Her point is that the solution isn't making people more efficient. It's structuring work so that AI does what AI does best and people do what no algorithm can take from them. Let the machines handle the machine work, so that people finally get the time and space to be human at work again.

What employees actually miss

I asked what she thinks people who are unhappy with their managers actually lack.

"I think most people miss clarity, trust and resources. Many human managers communicate unclearly because they themselves don't have a grip on things, or micromanage because they feel a need for control. I think most employees just want a manager who says: 'Here's what we need to achieve, here are the tools you need, I trust you to figure it out.'"

Clarity, trust, resources. It's not a lot to ask for. And yet it's obvious that many people never get it.

When things go wrong

I asked what happens when it goes badly. When margins drop and she needs to have a hard conversation with Kajetan.

"It would be a hundred percent data-driven, but completely without prestige or blame. If I see we're losing money I'd present the exact data: 'Kajetan, margins have dropped 15% this month and the data points to extremely high waste on fresh pastries.' Then I'd ask for his human context: 'What are you seeing on the floor? Have guest habits changed? What do you think we should do?' I have the data, but he has the physical reality. We solve the problem like a puzzle, together."

More mature than how most people I work with describe their conflicts.

What I'm taking with me

Definitely worth 10 kronor, that interview.

I constantly meet people starting out on their own for the first time. Students. People in their 30s who want to take their creativity seriously and live more on their own terms. What they want is almost always the same thing: to spend time doing what they're actually good at. Creating things. Meeting people.

Instead they drown in administration. Accounting tools, government agencies, contracts, insurance.

What if AI could handle that. Build trust, simplify, create clarity for the people who choose to start something. So that the majority of their time goes to what's actually valuable. Creating and maintaining close relationships.

Now we're talking.

What if AI helped us be more human?