I recently moved into a new apartment on the edge of Zürich, tucked somewhere between the alpine forests and lakes. It has been an exciting time - but also a surprisingly effective reminder on just how many decisions are required to exist as a functional adult.
I recently moved into a new apartment on the edge of Zürich, tucked somewhere between the alpine forests and lakes. It has been an exciting time - but also a surprisingly effective reminder on just how many decisions are required to exist as a functional adult.
Beds. Wardrobes. Plants I will almost certainly forget to water and light fixtures which, despite a qualification in electrical engineering, I have discovered is less about technical competence and more about patience, ladders and wondering whether living by candlelight would actually be so bad.
Yet, looming above all other choices is the high stakes game of interior design: choosing the sofa.
Because a sofa is not just another piece of furniture. It is the gravitational centre of the living space. Everything quietly orbits around it.
It’s where evenings slow down. Where conversations stretch. Where you huddle up fighting off the flu or half-watching the next episode of Young Sherlock while pretending I had any intent to fold the pile of laundry sitting beside me.
And unlike most things you buy, you don’t casually replace a sofa next week if you change your mind. Which got me thinking about something slightly ridiculous:
What if I let AI choose it?
Not assist. Not recommend. Not generate options I politely ignore.
I mean fully outsource the decision.
In theory, AI should be excellent at this.
I could give it the room dimensions, lighting, colour palette, floor materials, budget and so forth. I could layer in browsing history, saved images, past purchases - everything that quietly signals my taste.
Within seconds it could scan thousands of sofas across the internet and determine the optimal one. Not just aesthetically compatible, but statistically aligned with my preferences.
The most me sofa possible. And honestly, it might do a better job than I would. Because my current strategy mostly consists of opening twelve browser tabs, comparing fabrics I don’t fully understand, and whispering things like “but does this feel like a 3pm Sunday sofa?”
The more I thought about it, the more reasonable it sounded.
After all, most of what I was doing manually - scrolling across tabs, comparing fabrics I don’t fully understand, and second-guessing was exactly the kind of thing AI is designed to do better. It doesn’t get overwhelmed. It doesn’t open tabs and forget why. It doesn’t get distracted by a sofa just because it looks like something from a film.
It just…optimises.
And if the goal is to find something that fits the space, matches my taste, and stays within budget, then objectively speaking, the algorithm should win. Which led to a slightly uncomfortable thought: what if the best version of my home isn’t something I choose - but something that’s calculated for me?
But the more I followed that idea, the stranger it started to feel. Because a sofa isn’t just something that fits a room. It’s something that suggests how life will happen in it. Once you look at it that way, the idea of optimisation starts to feel slightly incomplete. AI can learn what I tend to choose. It can identify patterns in colours, shapes, layouts. It can approximate my taste with surprising accuracy.
But a home isn’t always built from consistency. Sometimes it’s built from deviation. From the one chair that doesn’t match anything else but somehow works, to the lamp bought on impulse that ends up defining the entire room. Or choosing something slightly bolder than usual not because it fits a pattern, but because you’re tired of the pattern.
These aren’t decisions you arrive at logically. They’re decisions you grow into. And that’s where it gets interesting.
From a data perspective, these are anomalies.
From a human perspective, they’re the point.
They’re the moments where taste becomes identity, where that irrational choice makes you intrinsically happy and it’s precisely why it feels personal.
And once you see that, another question starts to emerge. Not whether AI can choose a good sofa. But whether it would choose the right kind of mistake. The slightly-too-large one that makes the room feel generous. The slightly-too-soft one that turns into an unofficial guest bed. The one you weren’t sure about at first - but end up loving six months later.
Because part of building a home is not just getting things right. It’s getting some things interestingly wrong.
Which brings an unexpected layer into it: risk. after all not all decisions carry the same weight.
If an AI recommends a bad film, the cost is two hours and mild disappointment. If it recommends a bad restaurant, you eat mediocre pasta and move on with your life.
But a sofa?
A sofa sits in your living room every day. It quietly judges you. It appears in every photo you take when friends visit. It becomes the place where you sit when you’re tired, when you’re happy, when you’re procrastinating.
If it’s wrong, it’s very wrong. Which makes the idea of handing that decision to an algorithm feel a bit like letting an algorithm choose your haircut. Technically possible. Emotionally terrifying. Some decisions are not just about accuracy, but as much about ownership.
What surprised me most in all of this is how meaningful the act of choosing actually feels.
Objectively, it shouldn’t be. It’s a piece of furniture. One of many decisions in a long list of things required to make an apartment functional. And yet, it doesn’t feel trivial.
A home is one of the few environments we get to shape entirely ourselves. Every object carries a small signal of taste, of identity, of how we imagine our lives unfolding. You walk into the room and think: yes, this is my space. Handing that decision entirely to an algorithm feels efficient - but also slightly like skipping a chapter in the story. Because choosing the sofa is part of creating the home. And that’s the part that feels difficult to outsource. Not the analysis. Not the optimisation. Not even the outcome - but the process.
Choosing is not just a means to an end, it’s part of how that sofa becomes meaningful in the first place.
So here I am, still sitting with the thought experiment:
Would I actually let AI choose my sofa?
Part of me is curious. It would be fascinating to see what the algorithm concludes when given everything it could possibly know about my preferences. Another part of me suspects that, even if the AI chose the objectively perfect sofa, I might still hesitate before clicking buy.
Not because I distrust the technology. But because somewhere between comfort ratings, reviews, colour matching and predictive modelling, there’s a human layer that’s harder to quantify. The part where a piece of furniture stops being a product and starts becoming part of a life.
And that leaves me with a question I’m still not sure how to answer.
If an AI could choose your next sofa for you…would you let it?
If you’re building something ambitious, I’d love to be part of it. Reach out to chat about your next product or growth challenge.
