Henrik Werdelin
Entrepreneur and startup studio founder who cofounded BARK and Prehype, the studio behind companies including Ro and Audos.
“If I said to an LLM, what is the next business that BarkBox should do? I bet you it'd say it should make a cat box, right? Because it is the most predictable idea you could throw at it… And so we went to do an airline, which is probably also a bad idea, but at least a different battle. And it's more original.”
Mostly Human with Laurie Segall · Tech Strategy & Big Tech · November 2026
Context
Asked what 'original' means when everyone has access to the same models, Werdelin, a BarkBox co-founder, gives his own example of what a model would and wouldn't suggest. His point is that models default to the most predictable idea.
Transcript
…pharma, if you're thinking about finance or defense, I think a lot of all these big businesses that feel very certain that they can just keep doing what they did yesterday is going to have a surprise when the AI entrepreneur workforce kind of come in and try to nibble of their good businesses.
…a lot of this. And her name is Sandra Matts. And she was talking about how AI, if we're not careful, makes us all a little more similar. It suggests safer choices. It guides us towards kind of the middle, taking out the special sauce if we're not careful. So how do we all put on our Ironman suit?
So I think, so firstly, thank you so much for the nice compliment. I'm not a super naive person. I also realize that what we're trying to do has endless amount of flaws and it will be very difficult. I do think that what we learned with social media is that the best way to predict the future is to make it. And we have to take, we have to be very, we have to be very serious about what we're building because obviously we've done a few things that wasn't just great. And so I think that we have a lack of originality with what is the good case for AI. I think a lot of us very fast go to places in our brain of like, what does it look like when it goes wrong? And I think it's quite easy to kind of see all those kind of paths and imagine those futures. I think what we as people who like technologies and who love newness. need to become much better at is like, okay, let's just come up with cool shit that will kind of make this be awesome for a lot of people. And I think I'm definitely putting my hand up and saying, I'm going to give that a go and we'll see if it works out. Now, back to the original…
Right. I love this idea that actually the human is the special sauce, right? If you're able to be original and come up with those ideas. And I like that idea of the human in the loop. I'm very curious when we talk about BarkBox and how AI would have told you to do catbox, but you decided to create…
I don't know what would be right and what would be wrong. I think, you know, my solution for that specific problem was I think other people have sought to build a cat box and that for many reasons have seemed not to be a good solution. Like our friends over at Harry's tried to do that at one point…
Well, it is interesting. It's because AI does. It's trained on data of what existed. And if you want to create a company that answers your question, oh, God, it sucks that, you know, fill in the blank. There's a good chance that that doesn't exist yet. And so using AI as an infrastructure just to…
More from Mostly Human with Laurie Segall
