UX for AI

EP 84. Redefining UX Design: A Deep Dive into Canvas and Workflows in AI-Powered App Design with Jesper Bylund, Ex-Head of Design at n8n

• Bonanza Studios

Share episode

Send us a text

In this episode of UX For AI, host Behrad Mirafshar sits down with Jesper Bylund, a freelance designer and former Head of User Experience at n8n, to dive deep into the fascinating world of UX design, artificial intelligence, and the no-code revolution. Jesper brings a wealth of experience from his time at n8n, a leading automation platform that has redefined how businesses and individuals approach workflow automation. Together, Behrad and Jesper explore the challenges and opportunities at the intersection of UX and AI, offering valuable insights for designers, creatives, and tech enthusiasts alike.

The conversation kicks off with Jesper reflecting on his journey at n8n, where he played a pivotal role in shaping the user experience of a tool that has become a sensation in the no-code space. He shares the story of how n8n evolved from a side project into a powerful platform that allows users to visualize and automate complex workflows through a node-based interface. Jesper explains how the team navigated the challenges of designing for such a versatile tool, and how they integrated AI capabilities to enhance its functionality. He also touches on the importance of rapid iteration and experimentation, especially when working with emerging technologies like AI.

As the discussion unfolds, Jesper and Behrad delve into the broader implications of AI for the creative industry. Jesper emphasizes that AI is not a replacement for human creativity but rather a tool that can amplify it. He discusses how AI can handle repetitive tasks, freeing up designers to focus on higher-level problem-solving and innovation. However, he also cautions against the pitfalls of over-reliance on AI, stressing the importance of maintaining a human touch in design. The conversation takes an interesting turn as Jesper shares his thoughts on the future of AI-driven design tools, predicting that we’ll see an explosion of creativity as more people gain access to powerful, easy-to-use tools.

One of the most compelling parts of the episode is Jesper’s explanation of how AI and no-code tools are democratizing technology. He talks about how platforms like n8n are empowering non-technical users to build complex automations and even entire applications without writing a single line of code. This, he argues, is leveling the playing field and enabling a new wave of innovation. Jesper also shares his excitement about the potential for AI to revolutionize research and development, citing his work on Find Insight, a research tool that leverages AI to conduct user interviews at scale.

Throughout the episode, Behrad and Jesper touch on a range of thought-provoking topics, from the philosophical underpinnings of the scientific method to the importance of questioning assumptions in design. Jesper shares his skepticism about the concept of “validation” in research, arguing that true progress comes from disproving hypotheses rather than confirming them. This leads to a fascinating discussion about how designers can adopt a more critical and experimental mindset in their work.

The episode wraps up with Jesper offering advice for designers and creatives navigating the AI revolution. He encourages listeners to embrace the uncertainty and rapid pace of change, urging them to experiment fearlessly and stay curious. 

Tune in to UX For AI for a deep dive into designing the future.

You can find Jesper here:

Interested in joining the podcast? DM Behrad on LinkedIn:
https://www.linkedin.com/in/behradmirafshar/

This podcast is made by Bonanza Studios, Germany’s Premier Digital Design Studio:
https://www.bonanza-studios.com/

00:00:00:00 - 00:00:30:04
Speaker 1
Welcome to UX for AI, the podcast where we explore the intersection of user experience design and artificial intelligence. Today we're joined by Jasper B learned, a seasoned freelancer and former head of User experience at Nate and a groundbreaking automation platform. Jasper shares his insights on designing for AI, the rapid evolution of no code tools, and how creatives can thrive in the age of AI driven innovation.

00:00:30:06 - 00:00:33:24
Speaker 1
Let's dive in.

00:00:34:01 - 00:00:36:02
Speaker 2
Deep breath, deep breath.

00:00:36:03 - 00:00:38:08
Speaker 3
Now we are here. So, Jasper.

00:00:38:10 - 00:00:39:08
Speaker 2
Yes.

00:00:39:10 - 00:00:42:03
Speaker 3
I gotta tell you, man, I'm in love with Nathan.

00:00:42:05 - 00:00:44:04
Speaker 2
Yeah, I can totally see that.

00:00:44:06 - 00:00:53:03
Speaker 3
I, I dream about it. I think about it, I it's just becoming my obsession. What's what's happening? Tell me you designed it.

00:00:53:05 - 00:01:09:24
Speaker 2
I mean, it's I think this is what happens to everyone when they start using anything first. They don't get it first. They look at it and they think like, what is this? And then after a little while, something clicks and suddenly you realize you can do everything with it. And after that, like, it's it's that tool that people want to use.

00:01:10:01 - 00:01:26:10
Speaker 2
I don't know if you saw this, but this last week on LinkedIn, there was this. Someone posted a graph of like which, NoCo tools people are searching for, and it then just took off last year completely. And it's like demolishing all of the competition, like, super quickly. I have no idea why.

00:01:26:13 - 00:01:52:14
Speaker 3
You know how obsessed I am. Like some people have, people have different fetishes. Some people, you know, that's why porn industries, 30% of traffic of the entire internet, you got people obsess on, people obsessed with me. Sports cards. There's a fish tool that it's massive. Generally, if people go fishing in different places in the you obsess it, you name it.

00:01:52:14 - 00:02:01:14
Speaker 3
There is obsession, my obsession. Here I go. And and I train my algorithm on the show me people messing around with new.

00:02:01:16 - 00:02:06:22
Speaker 2
That must be a really good feed. I haven't actually tried that. I should check that out. That could be very cool.

00:02:06:24 - 00:02:38:08
Speaker 3
It's very hard to train the algorithm. Of course, I think that's something that they got to really get behind it, because is still anchored towards giving you the hard hitting, most engaging content, which is totally which is which is about the type of content we all know about extreme left and right politics. And and for me, as a creator like David Lynch rape, I cried that night when he was I started watching, I started drinking wine, and I'm watching a lot of YouTube from him.

00:02:38:10 - 00:02:51:14
Speaker 3
He has a fantastic, fantastic 50 minutes piece on transcendental. I don't know if I pronounce this properly transcendental meditation, which prompted me to do meditation. He says in on the topic.

00:02:51:16 - 00:03:00:09
Speaker 2
Oh that's awesome. We could totally get into that. I've been into meditation for the last like ten years, and it solved my insomnia, which has been plaguing me my entire life. Oh yeah, pretty cool.

00:03:00:11 - 00:03:01:00
Speaker 3
Tell me about it.

00:03:01:02 - 00:03:10:16
Speaker 2
Well, so I've had insomnia my entire life. I've started sleeping. Yeah. Since when? I was, like, six, seven years old. I remember having issues sleeping.

00:03:10:16 - 00:03:14:03
Speaker 3
Why? Your six year old baby.

00:03:14:04 - 00:03:35:08
Speaker 2
And I was the youngest of four brothers. So big family, very secure upbringing. Like, absolutely no issues, no stress, no nothing like that. Nothing. And still couldn't sleep. Had no freaking clue why. Went to doctors for a decade plus. Tried all of the tips like I've done sleep studies with, you know, sleeping with the electrodes in the lab.

00:03:35:08 - 00:03:52:05
Speaker 2
Everything. Nothing. Nothing was an issue at all. Nothing about the environment was bad. Everything totally worked. Then I was like 27. I picked up headspace and I started trying that out. At the time, I thought it was like, oh, this is the hack that's going to make me smart, because that's always been my goal. Anyway, that didn't work.

00:03:52:05 - 00:04:01:03
Speaker 2
But, what did end up happening was that I realized that I have absolutely no trouble sleeping. I was just getting stressed about not being able to fall asleep so I couldn't fall asleep.

00:04:01:07 - 00:04:05:05
Speaker 3
Oh, are you the youngest of four or middle or youngest.

00:04:05:07 - 00:04:06:20
Speaker 2
Of the youngest?

00:04:06:22 - 00:04:11:04
Speaker 3
Oh, wow. So the meditation helps you to relax.

00:04:11:04 - 00:04:31:00
Speaker 2
And it was like. It was like day and night, literally. Like I went through the first ten days of this, and I remember sitting at my office back then and I was like, I had a, I had a an epiphany or like a small insight of like, wow, I'm very rarely actually aware of what I'm thinking. And after that, I've slept soundly almost every night.

00:04:31:00 - 00:04:36:22
Speaker 2
I still get like, short periods where I have trouble sleeping, but it's very, very uncommon now.

00:04:37:03 - 00:04:37:19
Speaker 3
Wow.

00:04:37:20 - 00:04:49:00
Speaker 2
It was a it was an absolutely crazy difference for me. So, you know, you know, what happens when that stuff like that happens? You become like a, almost like an extremist sort of proposing meditation to everyone.

00:04:49:00 - 00:05:09:02
Speaker 3
Yeah, I, I have a very long history with meditation. So just to give you a glimpse of how dedicate that I've been, it's on and off with meditation, you know, you know, you sometimes you go really hard and sometimes you don't do it for a year or two. It's it's everyone I talk to. That's our story. You're not saying anything new.

00:05:09:02 - 00:05:37:03
Speaker 3
Everyone's is is like a roller coaster. But I went to India in the middle of India to do a Vipassana. Wow. It was. It was not so. So we basically we landed in Mumbai. It was for two months. So my my girlfriend, wife at the time, girlfriend at the time, wife now we were like, okay, we want to start of our Indian, India trip with the Vipassana.

00:05:37:05 - 00:05:49:24
Speaker 3
Where is the where is this? Basically the the camp is not a camp, but the Zen center. Yeah. That is closest and offered the soonest meditation for it.

00:05:50:01 - 00:05:51:10
Speaker 2
Yeah. How long were you in there?

00:05:51:11 - 00:05:51:24
Speaker 3
Ten days.

00:05:51:24 - 00:05:53:16
Speaker 2
Jesus.

00:05:53:18 - 00:05:54:22
Speaker 3
I know.

00:05:54:24 - 00:06:00:21
Speaker 2
It spontaneously. You just thought it was a great idea to just do ten, ten days of silent meditation.

00:06:01:02 - 00:06:32:04
Speaker 3
Ten days a is in a in a in the basically Zen center that is that was underdeveloped. So basically we were sitting in a, in this like, mosque equivalent of monastery that yeah, they, it didn't have proper roof. So I think crazy. Let me just tell you one example. So it was, it was not even it was is it was in a wasn't in Mumbai or like one of those like a, you know, advanced cities.

00:06:32:04 - 00:07:00:22
Speaker 3
It was like in the middle of India. So we just had to go in the middle of India and in a rural, in the rural neighborhood of that city, so around five. So we would sit from four, 5:30 a.m. to literally, what, 839 9:09 p.m. every day, every day, with the lunch in between which was the spiciest food ever.

00:07:00:24 - 00:07:04:06
Speaker 3
My my my God.

00:07:04:08 - 00:07:10:08
Speaker 2
And you could you couldn't talk to the people. You could tell them like, please, no, I had to, I.

00:07:10:08 - 00:07:17:16
Speaker 3
Had to find a chef was like, can you give me something? So it's sort of give me boiled, which vegetables with rice every day.

00:07:17:20 - 00:07:19:08
Speaker 2
So that's good though. Yeah.

00:07:19:08 - 00:07:27:09
Speaker 3
So they were, they were accommodating. Yeah. So we were the only English speaking people in that we partner. Oh my.

00:07:27:09 - 00:07:27:22
Speaker 2
God.

00:07:27:24 - 00:07:39:07
Speaker 3
Because it was in the middle of nowhere. They were not expecting us. So why are you guys here. So it's meditation. That's what you're here. We just want to try to pass for the,

00:07:39:09 - 00:07:43:19
Speaker 2
So we figured a ten day silence retreat would be a good start. Well, yeah.

00:07:43:21 - 00:08:14:17
Speaker 3
Yeah. So a lot. So, yes. So from 5 p.m. to 7:37 p.m., we would hear the mosque. Allahu akbar. Allo Akbar. The then Buddhist, they had their own drum rolls. They have then you got Hindus, you got like basically it was full on voice and noise from five people to 7 p.m. being up from 530, meditating and you had to really keep your focus like bunker.

00:08:14:19 - 00:08:18:21
Speaker 2
Wow. So, so like would you do it again? Have you done it again?

00:08:18:23 - 00:08:43:13
Speaker 3
So here's the thing. After we've we've done the meditation, the Vipassana, we, got to the town to hop on the train and go to south and, and before going, one of the people in the, in the we pass on a it's like, no, you come to me, you come with me, I take you to my family, you are going to give you a lunch.

00:08:43:15 - 00:09:10:24
Speaker 3
Anyways, after that, we also like straight up something great and we ended up becoming, it's sort of like a, you know, in the advertising, you got people. We got. We got the became the face of a three store. Three store shop. That's, really luxury clothes because they were like, you come here, we take some photo, we want to put it on on Instagram.

00:09:11:00 - 00:09:12:12
Speaker 3
Sure, but all of them.

00:09:12:12 - 00:09:16:14
Speaker 2
Were like, so you you were like the influencers, the the face of the.

00:09:16:16 - 00:09:19:03
Speaker 3
Crazy. Crazy.

00:09:19:05 - 00:09:20:14
Speaker 2
It's.

00:09:20:16 - 00:09:48:19
Speaker 3
So we sat in the chair and after all of this, I could hear what's happening in wagon on the other side of the train. It was so clear. Yeah, taste was clear. Hearing was clear. I was in a constant, literally Zen state of Zen. I was experiencing it for a couple of weeks. The level of attention I've had after that ten days is not humanly possible.

00:09:48:20 - 00:10:06:06
Speaker 3
Whoa. We are right now in a normal state of living. We we cannot understand how much of a potential a human has just to be personal for ten days. And you realize I. I really missed it, but, I'm not doing you you personal meditation anymore. I'm. I'm listening.

00:10:06:08 - 00:10:21:21
Speaker 2
No, no, I have a I have a bunch of friends who've done this, and none of them stick to it. After a personal, you know, it all go like, yeah, from now on, I'm going to do an hour a day every day. The world's going to do at least one hour per day at least. And I don't think anyone has done it.

00:10:22:02 - 00:10:24:07
Speaker 2
So I do 15 minutes per day.

00:10:24:09 - 00:10:27:21
Speaker 3
That's that's good. Yeah. I do 15 minutes. So I think 15 minutes.

00:10:27:21 - 00:10:44:02
Speaker 2
It starts in this little like trade off like you, you can do 15 minutes at it's fine. You'll be a little calmer. You'll be a little bit more perceptive. But you're not going to go like listening to rats on the other side of the street or something weird. But you, like, have you studied any neurology? Do you know why this happens?

00:10:44:04 - 00:10:44:11
Speaker 3
No.

00:10:44:15 - 00:11:19:11
Speaker 2
This is super, super fascinating. Okay, so like so the brain is a bag of fat, right? And what happens when you think is that you activate neural pathways through the fat and the ones that you use the most, they start getting encapsulated in this gooey matter that makes those pathways more efficient than the ones next to them. So the the common like metaphor for this is imagine you're standing in front of a field of fresh snow, and if you like, drag something through it or like you, you go skiing through it first.

00:11:19:11 - 00:11:42:04
Speaker 2
The tracks are very shallow, but if you go for the same route several times, the tracks deepen and you get these like really thick tracks that are hard to step out. Okay, so having said that, like, this is what you're doing constantly with your brain. You're basically training it to be more in tuned or be like more, attentive to certain things, better at thinking, certain patterns of thought.

00:11:42:06 - 00:12:05:15
Speaker 2
And the reason for this is that we take in a lot more information that we can actually perceive or handle or like process a lot more information. So we need something to filter this out. So this is what's training your mind to being more efficient. It's filtering out everything else. But this is of course like you can see the fly in this right.

00:12:05:15 - 00:12:28:09
Speaker 2
If the pants become too deep, suddenly you're stuck and you can't get out of it. So you end up having this like recurring rumination, thoughts like, anxiety, for instance. This is quite common today. And when you meditate, what you're doing is you're like loosening up the pathways a little bit. So you're making your brain more plastic by basically allowing it to rest.

00:12:28:11 - 00:12:54:11
Speaker 2
This is super, super like simplification, but it's approximately what's going on. So you're basically adding snow on the field of snow. So the tracks are less steep. And you can do more things with the brain. This is exactly the same thing that's happening when you drop acid. But the snow is way more aggressive when you drop acid. So suddenly you're just like, you're very, very creative all of a sudden and your pathways are like, you could go anywhere with these things.

00:12:54:11 - 00:12:58:00
Speaker 3
It's like bulldozer ING through the path. Yeah.

00:12:58:02 - 00:13:22:14
Speaker 2
Pretty much pretty much like the interesting thing about this, right. Is that what you just said, like this isn't humanly possible to be this perceptive. That's absolutely true. Like, only children can do that. Because if you do this as an adult, you're just going to be massively freaking distracted constantly. And you can sort of see this with like Buddhist monks, like they seem very, very calm all the time.

00:13:22:14 - 00:13:50:14
Speaker 2
Who else like what normal person seems that calm all the time. It's someone who has like brain damage. It's because they're like they're functionally equivalent to someone who doesn't really care. And they're not apathetic. They're happy. But they still don't really care. So is this a is this a better or worse situation? The one you're in? I don't know, but it's it's really fascinating to think about this and like think about what this means because it's basically training the algorithm.

00:13:50:14 - 00:13:56:07
Speaker 2
What like you said you were doing with X were basically training the algorithm constantly, like how do you want your mind to work?

00:13:56:12 - 00:14:08:13
Speaker 3
It's it's huge. I mean, I was expecting that our conversation to derail very quickly. I wasn't expecting it that much, but I think it's.

00:14:08:13 - 00:14:09:15
Speaker 2
Oh, sorry. Yeah.

00:14:09:17 - 00:14:28:21
Speaker 3
No that's fine, I love it, I love it. I think, we're going to talk about the AI at some point. Guys. Yeah, he does a lot of it. He's he's doing a lot of it. AI is behind in it. And so we're going to get to a new patient with us to.

00:14:28:23 - 00:14:57:12
Speaker 3
Oh there's but I think, you know, there is you know you know, there is there is a lot of relevance to the theme of the podcast. The team is about the innovation, UX innovation. It's about design and creativity in the age of AI. And I think I think, yes. But I think all of us creatives, we are in this semi-state of panic, excitement, enchantment with AI.

00:14:57:12 - 00:15:19:20
Speaker 3
I feel like there is a special, a spiritual reckoning happening, and I see it in the posts that I'm making. I actually, I'm a very optimistic person. I'm a very positive person. I mean, my bring in my the way that my upbringing and the path to if I allow myself to be negative, I would have been broken down.

00:15:19:20 - 00:16:05:07
Speaker 3
So but when whenever I post a very positive, very positive post, I got 400 comments, 100,000 of you on the post. I thought that is very motivating and positive. The responses I see. I think I need to do a PhD thesis on the response is so interesting. So what I want to take where I'm going with this is that I think all the options on the table in the age of AI revolution for creative to stay centered, to stay focused, to not lose the creative juice because it's the pace of innovation is going to even accelerate even more than this deep seek.

00:16:05:11 - 00:16:06:01
Speaker 2
Yeah.

00:16:06:03 - 00:16:09:16
Speaker 3
Got introduced two weeks ago. My god. Yeah.

00:16:09:18 - 00:16:14:21
Speaker 2
That's right. And deep learning from OpenAI is one week old.

00:16:14:23 - 00:16:15:12
Speaker 3
That's great.

00:16:15:12 - 00:16:18:05
Speaker 2
And by the time this ships, there's probably going to be a new thing.

00:16:18:07 - 00:16:34:06
Speaker 3
I think deep learning, the ChatGPT, deep learning is going to make a lot of this survey. Research consultancy that charge 20,000 to create a thousand pages report out of business very soon.

00:16:34:08 - 00:16:40:13
Speaker 2
Yeah, yeah, totally. Like things like I wouldn't want to be at McKinsey right now.

00:16:40:15 - 00:16:51:21
Speaker 3
Maybe expand because you and I are so deep in that we we see why maybe for for. Yeah. Why do you think McKinsey there is not that another pathway for McKinsey in the future.

00:16:51:21 - 00:17:12:04
Speaker 2
So a lot of like the basic work that huge consultancies like McKinsey do is they do like foundational research about markets. So what they do is they go out and they just find a bunch of data points, and then they use this for analysis with their clients. And they charge a lot of money for this because nobody else is really doing it.

00:17:12:04 - 00:17:41:17
Speaker 2
And it needs to be specifically tailored to your client or your marketplace or whatever. But, now with with deep research, OpenAI is basically allowing you to do exactly the same thing, only on digital assets. But still, most things today are digital, for nothing like it's down to a few dollars. I think currently you have to be one of the $200, subscribers to get access to it, but that's just because they're still rolling it out.

00:17:41:19 - 00:17:57:10
Speaker 2
The the API cost of running this is going to be negligible. So for something like 2030 bucks, you can probably get a report that would cost McKinsey 20 €30,000 to produce, which is just absolutely crazy.

00:17:57:10 - 00:17:59:10
Speaker 3
If not 200 300,000.

00:17:59:10 - 00:18:17:14
Speaker 2
Oh yeah. Yeah. Charge to the client. Definitely. So all of these basic research projects suddenly become accessible to everyone, and it's no longer a moat for big business. And it's no longer like a, like a point of leverage for a consultancy to have this, like, give a quick overview.

00:18:17:16 - 00:18:43:16
Speaker 3
Yeah. No. It's fantastic. I think we actually it's interesting because, you know, we talking to to big enterprise and they and I know the global head of marketing of do you know we've done project for years. He said like Bharat we got a code for a survey, research 1000 pages for 35,000. What do you think? I didn't say I could do it for 200, but I would, I think like.

00:18:43:18 - 00:19:11:01
Speaker 2
You should up. You should, you shouldn't. You should have just to see what their reaction was like. We have to remember this, this old, sci fi quote. Right? Like the future is already here. It's just not widely distributed yet. And this is going to be super true for AI. And like, I think for the next ten, 15 years, we're going to have this conversation over and over and over again with clients all over the world, because everyone is going to be like, no, but this specific thing you can't do with AI?

00:19:11:01 - 00:19:23:10
Speaker 2
Yeah, you totally can. You totally can. But I don't think, like you said, there was like a bit of an existential anxiety in you. For as a creative about AI, I don't think you need to worry like I'm not excited.

00:19:23:10 - 00:19:25:09
Speaker 3
Why? Why are you excited?

00:19:25:10 - 00:19:31:08
Speaker 2
Maybe because this. Okay, so this is going to be a complete tangent. No, I don't know.

00:19:31:08 - 00:19:33:23
Speaker 3
Maybe I'm with you. Go wherever you want to go.

00:19:34:04 - 00:20:00:23
Speaker 2
Every time there's a new, powerful tool released. Every time what happens, there's an explosion of creativity. Like every time the the cost of, like, cameras get lower. What happens there are better movies, there are better YouTube clips. Everyone is just empowered to do more things. What is AI? It's just a way better tool. And like, I think a lot of people are worried that AI is going to replace you because it's going to be doing your thinking.

00:20:00:23 - 00:20:20:09
Speaker 2
And it's it's not it's not thinking, it's not reasoning, it's not actually creating anything. It just has a shit ton of information encapsulated in its neural net, and it's accessible for you. But you still have to ask the right questions. You still have to lead it down a path. You still have to connect this to some sort of actual customer value.

00:20:20:09 - 00:20:38:01
Speaker 2
Like, I'm not I'm not worried at all. I, I'm look, I look forward to the world where I don't ever touch code. I just tell the computer what I wanted to do. I don't ever, like, think about the day. You don't have to draw rectangles anymore because you can just ask it to just, like, make me a user flow that has like these steps.

00:20:38:02 - 00:20:39:07
Speaker 2
That would be awesome.

00:20:39:09 - 00:20:52:12
Speaker 3
Awesome. I think you've said it. I think that's why creatives and designers need to be so bullish and optimistic, because liberation is here. You don't have to depend on developers to do your thing anymore.

00:20:52:14 - 00:21:11:20
Speaker 2
Yes, yes. Just that thing like that. Is that is a mind blowing statement. Think about like if designers and creatives can just go out and do their own apps you like because this is the world right now. Like if you go on Replit right now, you can launch an app today, it might not be awesome, but like in six months, it probably can be.

00:21:11:24 - 00:21:35:24
Speaker 2
Yeah. And if every creative can do that, I want to live in this world. Like, this is going to be so interesting because we're going to have an explosion of people who are just be able to suddenly do a lot of things that cost like €2 million. Three years ago, like when I was at Volocopter three years ago, we we built like a customer app for months, like we had we had an external developer team and we set up our own internal developer team.

00:21:35:24 - 00:22:06:21
Speaker 2
And like we struggled and argued about the the flow of the app. Today, I can generate that myself. Like I could replace 13 of them in productivity. I can't replace their knowledge. Right? So then that team is just able to do 13 times more stuff. But I could try this idea now, like instead of instead of going through the whole thing with 13 different people in 13 different wheels dragging in all sorts of directions, I could just sit down and try it and go to the team and be like, hey guys, look at this.

00:22:06:21 - 00:22:17:03
Speaker 2
This is an app. You can download it on your phone and try it. What do you think is this solution? What we're going for now like that is so much more interesting than the current use case. Yeah.

00:22:17:03 - 00:22:28:15
Speaker 3
I'm, I'm. Hey, I'm getting goosebumps. I think I, you know, for me, if I want to dumb it down, it's like a Tesla robot equivalent of a horse hooked up to a nuclear power plant.

00:22:28:17 - 00:22:30:21
Speaker 2
And you need to, if you want to boil it down.

00:22:30:21 - 00:22:36:03
Speaker 3
Yeah. And you need to write it like how you write it is up to you.

00:22:36:05 - 00:22:54:15
Speaker 2
I think this is hard to talk about, especially for creators, because there's so little understanding of what the model actually does. Right? We can't look into it. We don't know what it's actually doing. So it feels like it's magic. And magic is scary. Magic and power is always scary. So we look at it and we're like worried about the worst possible outcomes.

00:22:54:17 - 00:23:17:10
Speaker 2
Yeah, that's our creativity working against us because the AI is still just going to do what you query it to because it's not it's not active. It doesn't do anything. It's just a database more or less. Yeah. And I think like one of the things we have to remember is what this like or how I think about what this is helps me sort of figure out what I should be doing with it and what I should not be doing with it.

00:23:17:10 - 00:23:40:09
Speaker 2
So how I think about this is taking it back to the brain thing we talked about a few minutes ago, like this is this is a massive brain who is out on the internet, and it's just memorizing all of these things. It's just reading everything, watching every video, looking at everything, and it's memorizing it, but it doesn't understand what it's looking at because there's no reasoning.

00:23:40:09 - 00:23:58:16
Speaker 2
Right. So if if you and I go on YouTube and we watch like a video about, would, you know, and then we, then we have to like, really, we have to, like, remodel our kitchen or something. Then we can go, like, oh, I could use wood for this. That would be excellent. Because wood, like, bends in this way.

00:23:58:18 - 00:24:21:17
Speaker 2
That would be awesome. And I can never do that unless somebody wrote it down. Because when we're doing that, what we're doing is we're taking two facts to separate facts, and we're thinking about an explanation for why these things could work together, like it's a creative problem. AI does not do this. You can try it yourself by getting it to do math or something.

00:24:21:17 - 00:24:33:07
Speaker 2
It's very, very easy for it to just go completely wild and start hallucinating. It's memorizing all of these things, but it's not. It doesn't understand them. And it's great for us because we can just do less of the boring stuff.

00:24:33:09 - 00:25:18:19
Speaker 3
100%. I mean, right now, I think, is the age that creative with, unique personality, character, unique tastes, folks that are more in tune with their with their calling, they, they're doing homework of personal development and they're becoming sharper, more aware and conscious individuals are going to actually becoming more in-demand, because that's exactly the type of personality traits that I cannot acquire the taste, creativity, gut feeling, sixth sense, whatever you want to name it and label it a I cannot obtain this.

00:25:18:19 - 00:25:20:22
Speaker 3
It's basically orchestration.

00:25:20:22 - 00:25:22:02
Speaker 2
All of the tastes.

00:25:22:02 - 00:25:22:24
Speaker 3
So yeah.

00:25:23:01 - 00:25:25:12
Speaker 2
It's like the opposite of generalist.

00:25:25:12 - 00:25:52:05
Speaker 3
It's it's a basically it's weights and biases, backpropagation and gradient descent. That's. Yeah, that's that's what I, I, I, I'm here's the thing. What are we to say to all the creatives. If you want my $0.02, if you're afraid of something understand it first. Yes to the core of it. So don't read on X and LinkedIn. Some person wants your attention and talk about what I get.

00:25:52:05 - 00:26:14:01
Speaker 3
A book that is sufficiently sophisticated if you want to read a book. I'm reading a book called how AI works written by Data Scientist, and this is all you need to have a bit of engineering background to get it, but it's not complicated. Read and learn because that's what exactly happened to me in the Bitcoin era. Everyone had thought about it.

00:26:14:03 - 00:26:31:24
Speaker 3
You call it flood, right? Had fear about it was like, okay, I get it, let me read. So I read the Bitcoin standards. I was like, okay, this is what Bitcoin is about. It's great. It's a future. I'm going to go all in. Same thing with AI. In the middle of all this fear, I was like, okay, what is this thing called AI?

00:26:32:00 - 00:26:46:00
Speaker 3
Let me read right? It's really helping me to understand what AI is and what AI is not. I like, I think your explanation, it's very on point. It's generally the engine cannot do anything else.

00:26:46:02 - 00:27:09:10
Speaker 2
That's I think that's like I just got it from what you were saying, with tastes like it's having an opinion, having a viewpoint is what creative work is all about. Being a designer is all about being shaping the attention of something in a certain direction for a reason. And yeah, you can query AI to do this for you, but it doesn't do it because it's been trained on everything.

00:27:09:10 - 00:27:14:24
Speaker 2
So it's just going to be the blandest soup of boring gray stuff you could possibly think.

00:27:15:00 - 00:27:21:03
Speaker 3
Unless you're prompted to say, okay, go look at this direction, at this silo, at this specific.

00:27:21:03 - 00:27:46:15
Speaker 2
Top. Exactly. Remember, in the like when, I think it was Dall-E or Midjourney, maybe that came out first and, had no constraints, so you could just tell it to like redraw this in the style of Picasso and it could nail that. It was really, really good at that. But creating a new style requires you to do a ton of work, and you can do the work manually, or you can do it with AI.

00:27:46:20 - 00:28:04:02
Speaker 2
You still need to go through all of the steps, and for that style to then become something that I can use, you need to give it basically a style guideline. So it's not going to be it's not going to be interfering with the real deep work of design any time soon. No.

00:28:04:04 - 00:28:05:19
Speaker 3
Far away maybe.

00:28:05:22 - 00:28:10:09
Speaker 2
What are you doing now to leverage AI? What's like the best things you've learned recently?

00:28:10:11 - 00:28:34:03
Speaker 3
Oh, no, I would say, and I would love for you to go deep in it because you you were the design brain behind it. And I think it I owe it to the audience to give you a space without distracting you to talk about it. But, we are just rethinking everything. Really? Yes. But I think starting with our SEO, all of our is going to be automated.

00:28:34:05 - 00:28:56:23
Speaker 3
So we'll be using four different bots doing it. So that's our bit. I don't need the content marketer unless I'm proven wrong. Right. And again this is not about again, I'm not trying to replace humans, this client, but I'm running a business. I'm paying people salary. I have to make sure my family is going to have food on the table as well as my team.

00:28:56:23 - 00:29:18:22
Speaker 3
Right. I have to make a decision that works for me and for the business. Right? It all comes down to the business decision. I always it's not about me trying to. No, that's not a case. It's a how can I survive? That's all comes down to it. So that's on the marketing side. I'm big, big, big, big on automation.

00:29:18:22 - 00:29:38:22
Speaker 3
So for example, I want to here's, here's what I would like to build on. Knative I want to say I will. So I will build automation Engine, let's say get my latest LinkedIn posts based on all the criteria I introduce you for best case, best practices to write this, blog post. Write it, publish it on workflow.

00:29:39:00 - 00:29:41:01
Speaker 2
Oh, yes. Yes.

00:29:41:03 - 00:30:13:02
Speaker 3
So that's these are the things that I would like to, experiment. Then there is another track I would like to experiment is, like you said, a replit sports lovable. Whatever the case, fill the blank. I think there's a massive future over there. For the type of client that we are working with, we don't need we cannot use this kind of application because there are, like, enterprise grade bespoke software, you know, whatever the future that we are developing, we have to extend, we have to scale it to like 20, 30, 40, 50, 400,000 people.

00:30:13:02 - 00:30:23:12
Speaker 3
It doesn't work with these things. One thing that I'm really excited I want to do it is that command based prototyping.

00:30:23:14 - 00:30:24:16
Speaker 2
Yes. So yes.

00:30:24:18 - 00:30:43:24
Speaker 3
We have a very elaborate design system. And I told the team, why are we? Because basically they go the way our team design. So I want very big two of the sign up page. It's center aligned. No. For this client we need a two column for this one. We need this is just playing around the variables. Is that okay.

00:30:43:24 - 00:31:04:09
Speaker 3
You guys are doing this. Why don't we turn our design system into a vector database, feed this to AI and write the product requirements center of Dram. It's a there is a lot of apps already doing it. So this I'm not saying something revolutionary. I just want to see how we can make ourself nimble, lean and more agile with the AI.

00:31:04:13 - 00:31:23:07
Speaker 2
Less of the boring stuff. Yeah, that's like honestly, that's genius. That's exactly what Vercel did with, v0.dev. Right. It's basically taking their design theme, which is shad cms slash UI. They're basically taking that library and just generating a bunch of websites.

00:31:23:07 - 00:31:24:02
Speaker 3
Oh, that's what.

00:31:24:02 - 00:31:25:21
Speaker 2
It doesn't read anything. Yeah.

00:31:25:23 - 00:31:33:12
Speaker 3
I heard a lot of folks today just for the good by saying good bye to Figma. And they're doing everything on V0.

00:31:33:15 - 00:31:52:22
Speaker 2
Yeah, I'm not surprised. But the only boring thing about that is you get their style. So afterwards you kind of have to go style it. But honestly that's fine. Like if you if you're solving a real problem, if you're doing something that is actually genuinely creative, you can just create the function first and then you can like make it look the way you want.

00:31:52:22 - 00:32:11:10
Speaker 3
It to look. And also I would say like this kind of solution really works well with your practice, which I folks, I have the pleasure to sit and listen to Jesper keynote on how he goes. He doesn't like to call it validate how he goes, test ideas. I know, I'm just I want to be careful.

00:32:11:10 - 00:32:18:06
Speaker 2
I don't want I not just don't like, validate. I am a staunch enemy of the entire concept, too.

00:32:18:07 - 00:32:27:15
Speaker 3
Like like at least I give it like five years. Rundown on why you think validation is a crap or not. My word is not your word. Sorry, I don't want to put words.

00:32:27:15 - 00:32:29:05
Speaker 2
Validation is crap.

00:32:29:07 - 00:32:31:24
Speaker 3
Okay, what?

00:32:32:01 - 00:32:46:01
Speaker 2
Validation is pointless. So, a couple of years ago, I read a book, and, this book is called, The Beginning of Infinity by a physicist called David Deutsch. And in it, he explains like a ton of very interesting things.

00:32:46:04 - 00:32:49:03
Speaker 3
A lot of the video is fabric of reality. I'm reading.

00:32:49:05 - 00:32:56:19
Speaker 2
Yes, yes, that's I still haven't read that one yet, but it's, it's like the problem is he is so incredibly.

00:32:56:19 - 00:32:57:03
Speaker 3
Oh, my.

00:32:57:03 - 00:33:19:20
Speaker 2
God, these things that the books just take over your entire life. So I'm a little afraid of this. When I read the Beginning of Infinity, I had to, like, download the podcast that explains every chapter just because I wanted to really, really understand it. So I think I spent something like 200 hours listening to the podcast about the book, which I also read twice, because I felt like I needed to spend the time to understand it.

00:33:19:20 - 00:33:20:05
Speaker 3
Get it.

00:33:20:05 - 00:33:48:12
Speaker 2
Out. I might be a little dumb, but it's also a great book. So basically what he explains is the scientific method. And Popper, Karl Popper is the philosopher who sort of defined what the scientific method is, or sort of how it works. And what he realized was that what we, in a sense, try to do when we're validating something is we're trying to go out into the world and we're trying to find proof that our belief is true.

00:33:48:12 - 00:34:12:16
Speaker 2
So it's the the like philosophically, it's called the justifiable true belief system of thinking. And you can do this to literally anything and find proof for that. You can take any political slogan. You can take any like religious belief, you can go out and you can find proof that that is true. This is why it's useless because it doesn't take you anywhere, right?

00:34:12:16 - 00:34:44:13
Speaker 2
You can spend infinite time validating things, and what happens is you will eventually validate it. Unfortunately, this has absolutely no aspect. This has no connection to forward progress. This is no connection to any objective reality whatsoever. The things that do have a connection to actually making real progress is negative knowledge. So going out into the world and trying to disprove what you believe, let me give you some examples because this is starting to get vague okay.

00:34:44:13 - 00:35:09:07
Speaker 2
So let's let's take something directly from our business. Let's do like a signup flow or conversion of a sign up flow. Right. So if we believe something like, we have a sign of flow, it's two steps. You click a button because you're interested, you fill in your email and you press sign up and that's it. Currently, the sign up is like 2% of visitors and we don't know why we want it to be like ten.

00:35:09:09 - 00:35:36:21
Speaker 2
We have no clue why. So we start thinking like, this is going to be great. Like, we can we can figure out why people are not clicking the button. We can figure out, like, what is the, what is the reason that people do this? So we start like talking to people who've signed up and all of the people who've signed up say they signed up, because this wonderful message we were using on a store page led them to be like, yes, this is for me.

00:35:36:24 - 00:36:01:18
Speaker 2
So we start using this message in advertising and we start like pumping it up and putting it on the second page and whatever. And the sign up ratio increases a little bit and we're like, yes, validation. This was true. The reason this really worked was because of this message. Awesome. Unfortunately, this is not true. You could do this with literally anything and the stats would probably change.

00:36:01:20 - 00:36:27:15
Speaker 2
And unfortunately, in the case of a sign up, if you just push it to more people, it will increase just because there's more people in the funnel. So it was slowly, slowly increase a little bit. So if you start believing stuff like this, you are not wasting time and money on things that are not true. So what you should be doing is you should be figuring out why is it that 98% of the visitors are not clicking this?

00:36:27:17 - 00:36:28:19
Speaker 3
Love that.

00:36:28:21 - 00:36:48:16
Speaker 2
And you should be figuring this out by making it lower. You should get it lower. You should figure out what is wrong because there's always a problem to solve. And solving a problem means you're making progress. You're making an actual improvement. But if you're trying to make an improvement without first defining a problem, you're in this space now of just like throwing spaghetti at the wall.

00:36:48:17 - 00:37:09:01
Speaker 2
Anything at work, you have no clue if it's true or not. It's just like random random action. So it's just busywork and it's infinite and it goes nowhere. But if you're doing the other way, like it's going to be very, very clear, like if you remove something and it drops 50% like, oh wow, okay. We definitely cannot remove this.

00:37:09:02 - 00:37:30:19
Speaker 2
Like, okay, can we what if we change it to something completely different? Oh, it doesn't drop. Okay. So this is probably useless. Can we remove it? Yes we could. Okay, cool. See, we're now making real progress about realizing what it is the customer actually cares about. So it becomes a very, like, concrete way of figuring real things out.

00:37:30:21 - 00:37:55:17
Speaker 2
And the strange thing is like this applies to literally everything. And Popper believe that, this is true for how we actually construct our internal knowledge about the world. So he thought, this is this is like physically, how the brain actually creates information about the world and that this is the only way to do it. And the interesting thing about this is that this is based on the premise of criticism.

00:37:55:19 - 00:38:17:24
Speaker 2
So what you're doing is you're finding errors and there are always going to be errors. There's always going to be things to improve. And that's a good thing. And then you're trying to criticize the reason those errors are there. So you can improve the reasons. So you can create less less strong errors or like smaller errors. And this is an infinite process of just getting rid of more and more errors.

00:38:18:00 - 00:38:28:11
Speaker 2
And of course within that thought structure, you could criticize Popper in the same way, but that is very hard. You should try it if you if you dare, you know. Is that too vague or is that.

00:38:28:11 - 00:38:50:10
Speaker 3
You know it's so 100% I can what I can add to it is like whenever for when I twisted my ankle, I got out after it. So I had to be in the cast for six weeks. Then I got out on the streets. Oh, I would, I did start seeing more people that are on the crutches. So it's basically, yeah, it's sort of like touch upon validation.

00:38:50:10 - 00:39:20:08
Speaker 3
When you want to look at something, you find the pattern that confirms what you're trying to always that's always the case. And also another thing I could add is that, we, we hear these things that this saying that trust the science. So whenever I hear this is like, well, I did a course I to for my masters, I had to do a course on scientific methodology, and the entire premise of the book was about truth, questioning the science.

00:39:20:10 - 00:39:32:22
Speaker 2
Yeah. I was like, I love, I love that phrase. It's hilarious. It's like every scientist is just rolling over like, no, this is the complete opposite of the point of science.

00:39:32:22 - 00:40:01:04
Speaker 3
Yeah. And one thing that I loved from what you said, and I'm going to actually implement it, I don't know how to implement it, but I'm going to really be careful about it, is that if in your sign up page you see 2% conversions, you see 98% basically disengagement or no engagement, don't try to optimize the 2%. Try to look at the other half of the class that more than half of the class is the entire class 98%.

00:40:01:07 - 00:40:15:04
Speaker 3
How can you reduce 98%? Don't don't try to. Because look to your point, you can increase your ad budget and bump that 2% to 3% to 4%. But have you made any improvements you still like 90% short?

00:40:15:06 - 00:40:39:08
Speaker 2
Yeah, exactly. And like there are scenarios for conversion and for business in general where you should no longer be optimizing this. You should just push ad budget. Right? This happens. But the question has to be at the first place. Like do you need to learn something new, or do you just need to work harder? Because when you need to learn something new, you should be looking for ways to disprove what you believe.

00:40:39:10 - 00:41:09:23
Speaker 2
But if you don't need to learn something new, don't go doing research. It's just a waste of time. And validation is by definition, a waste of time. It's completely useless and we should just not be spending our time on this. Which is why I'm so frustrated when I go to all of these startup meetups. And when I talk to all of these, like, researchers that I meet because most of them don't know this, they do research and they end up having no results for years and years and years.

00:41:09:23 - 00:41:32:10
Speaker 2
The result is always, well, kinda, well, sorta. It's like 5050. No no no no no, that's not research. That's pretending to do research. Well, no, I shouldn't say pretending because that's unfair. It's not the doing research. Yeah, yeah, they're doing the best they can and they're making a real effort. The problem is, do you know about cargo cults?

00:41:32:10 - 00:42:02:05
Speaker 2
We talked about this last time, right? No. So in the South Pacific, after the Second World War, there were a bunch of tribes that were not super evolved yet or not evolved. They were not super high tech yet. And they'd watched all of these American fliers, build an airstrip, build a little air tower, sit in the air tower with with headsets on, and then a plane would land and it would they would get a ton of stuff like food and chocolate and all sorts of things.

00:42:02:07 - 00:42:29:24
Speaker 2
And when everybody left, when the armies left the South Pacific, some of them started actually reengaging this. They started making airstrips. They started building air towers. They started building headsets out of like bamboo and coconuts and stuff. And they would sit in the air towers speaking gibberish, waiting for the plane to land, because they didn't understand why the plane landed.

00:42:30:01 - 00:42:50:10
Speaker 2
And like, the thing is, they did this in in modern days, right? This is 60, 70 years ago. So there are photos of this. So this means that every single one of us is prone to do this when we don't understand something, we ape it, we mimic the motions and we go through the motions. And we're basically in a cargo cult.

00:42:50:10 - 00:42:58:05
Speaker 2
And I see this constantly. I used to do this for, for like the gym and I think, like, I'm sure I'm still doing this in certain areas.

00:42:58:05 - 00:43:03:16
Speaker 3
Of my life, but in the photos, yes. For it's crazy.

00:43:03:18 - 00:43:05:08
Speaker 2
It's bizarre. Right.

00:43:05:10 - 00:43:13:01
Speaker 3
They made a massive statue of it. Satellite dish. Yeah. Whoa.

00:43:13:03 - 00:43:36:15
Speaker 2
Yeah. A decent modern human. This is now right there. Kids have cell phones, so this isn't. This isn't something that happens to, like, you know, this isn't something that happens to chimpanzees. This is happening to us. We are prone to have this problem. We are prone to mimic things without understanding how it connects to reality. And we just go through the motions instead.

00:43:36:21 - 00:43:58:22
Speaker 2
And unfortunately, I think a lot of research is that I think a lot of research is just going through the motions. It's collecting all of these data points with no real attribution. It's doing all of this marketing and doing all of this development, but nobody actually understands what it's supposed to ship. It's like disconnected from the business value, but also from the user value.

00:43:59:02 - 00:44:20:18
Speaker 2
And it's disconnected from the mechanics of what we're actually doing. And I think like this is super legit a problem for jaws, right? When they come out of school and they start working, pretty much all of them have this problem where they're just trying to perform, but what they're actually doing is more of a performance than actually delivering.

00:44:20:20 - 00:44:36:08
Speaker 2
And it's our fault as seniors when we're not explaining how this connects to the real world for them, because they need to learn somehow, and they can only learn two ways, right? They can learn by failing to do it. Or we can help them understand this, this no other way.

00:44:36:08 - 00:45:00:21
Speaker 3
So, I would love on the podcast we have about so we have about community around 10,000 designers, meetup and webinar and email newsletter. I would love for you to if, if you have time to basically I've produced a webinar, you do a, presentation on doing research properly. Sure. I would love that. And then. Yeah.

00:45:00:21 - 00:45:09:11
Speaker 3
And then if you see the distractions, feel free to do a course about it. And, you know, so just talk about this off of the podcast.

00:45:09:13 - 00:45:40:05
Speaker 2
Sure. But like, this is, this is why I started finding site because I was, I was like then I was helping them with the, with our AI, integration. And we had this issue where we needed to learn a lot of things and we needed to learn them really, really fast. And we were super limited in scope. We had like we had three designers, we had two PMS, and we were calling people, calling them, calling them, calling them, trying to get them on the phone, like trying to get interviews booked and trying to find people who we could talk to about AI.

00:45:40:05 - 00:46:07:00
Speaker 2
And this was like two years ago, so nobody knew anything about AI. And we eventually, like we realized that we were just not learning fast enough because we couldn't talk to them fast enough. So the the founder, Yann, he basically said, I think this is the metaphor we're going to take. We're going to learn from, from this project, this open source project called learning to Ever going to integrate this because they're very parallel to how anything is structured.

00:46:07:02 - 00:46:25:00
Speaker 2
So we're just going to basically push push them together and we're going to do the same thing. And it's going to work really, really well because it's a similar model. And we looked at it and we were like, yeah, this is a really good bet. Completely unfounded from people we've talked to. And it turned out to be us.

00:46:25:00 - 00:46:42:06
Speaker 2
It turned out to be a really, really good bet. But like coming out of that, coming out of that project and like, I think I was there for six more months or something. And after that I still had this like gnawing feeling in me, like, why wasn't this fast enough? Because I had similar issues when I was at Volocopter before that.

00:46:42:06 - 00:46:57:15
Speaker 2
And I realized that, like, we need to interview people to learn, but we can't interview them fast enough, so we can't talk to people fast enough. Which is why I started finance. I thought I because I just wanted to leverage the time I had to talk to more people.

00:46:57:18 - 00:47:09:00
Speaker 3
Out of selfish reason, because we have a big research project coming up. What does Find Insight do for teams and companies and how is that different?

00:47:09:00 - 00:47:30:06
Speaker 2
So what Finisar is, is, is basically a research agent. So it replaces, user researchers calling up your users and doing interviews. I shouldn't say it replaces because it's not as good as a human being. So a researcher, we still do a better job. The difference is that finding. So I can do this to 10,000 people on a week.

00:47:30:06 - 00:47:48:18
Speaker 2
Whereas a researcher can do it to what, like five. So it's a leverage. It's a scale leverage difference to how many people you can talk to. And because it's a synchronous and a chat instead of an interview, you also reach a lot more people who just don't have time to schedule or feel like they don't have time to schedule an interview.

00:47:48:18 - 00:48:05:04
Speaker 2
So that's what it does in a nutshell. So what it looks like is that you as the, as the, researcher or the the, the person who wants to do research, you have a bunch of outcomes. You need to learn. I need to figure out why aren't they signing up? I need to figure out why aren't they using this, this feature?

00:48:05:04 - 00:48:22:04
Speaker 2
I need to figure out what other tools are they using. And you don't need to structure that into like a research question. You don't need to set a plan for that. You just give it about. These are the things that I want to know about this. Okay, great. Here's the link you should send to your users. And it does the communication for you.

00:48:22:05 - 00:48:24:15
Speaker 2
It does the research interview and you.

00:48:24:15 - 00:48:40:09
Speaker 3
Get the answer. So basically you just pass it on to your user base. Hey, talk to this part at your time whenever you want. However, you want the bathtub on the desk, whatever you want to do, it talks to you and fetch the information we need from you. Yeah. Wow.

00:48:40:09 - 00:49:04:18
Speaker 2
And because it's a because it's an alarm and the behind of it, it's smart enough to understand, like this user stressed I should back off. This is a vague answer. I should ask three follow up questions to make sure that it's super specific so you get really, really good results. Still not as good as a as a senior researcher, but I compared it to some junior research that I've done recently.

00:49:04:20 - 00:49:17:17
Speaker 2
And, it's definitely comparable to to junior researchers doing interviews. So I'm shocked at the level of quality of being able to get it to, so far, because it's still only very basically fine tuned.

00:49:17:19 - 00:49:28:11
Speaker 3
And do you offer also synthesize and reporting and maybe affinity diagraming or anything else as such, after the research campaign is over.

00:49:28:11 - 00:49:52:02
Speaker 2
So there's no such thing as over in Find Insight, you can watch the results come in in real time and oh fantastic. And it recalculates the, the executive summary in real time and what kind of results you're looking for. That is still something that I'm doing research on. So I'm still like building out the toolset for how we should interact with it.

00:49:52:04 - 00:50:07:23
Speaker 2
But currently there is a, there is a executive summary, which is, showing themes. It shows some of the most important quotes. It's just summaries per, per research outcome you want, and it gives you a, the full transcripts of every conversation.

00:50:08:00 - 00:50:12:13
Speaker 3
Oh that's brilliant. So you're also using it for your own project? Sue. That's great.

00:50:12:15 - 00:50:14:02
Speaker 2
I'm using it to build it.

00:50:14:04 - 00:50:23:12
Speaker 3
Yeah. I just told another founder. He's asked me the other day, hey, how do I know my product is ready for the market? I told him that when it could help you.

00:50:23:13 - 00:50:34:04
Speaker 2
That is a good rule of thumb. Yeah, I mean, when you're building something that is a tool, it either works or it doesn't, right? It either does help you or it doesn't help you.

00:50:34:05 - 00:50:36:09
Speaker 3
It's very binary. I get it, it is.

00:50:36:09 - 00:50:45:09
Speaker 2
It really is. And then after it does work, you can probably improve it a thousandfold, but it still has to start binary. So should we go back to edit N or where do you want to go from here.

00:50:45:09 - 00:50:51:14
Speaker 3
Yeah I mean it's about 60 minutes. We've been talking I could I could go on forever.

00:50:51:16 - 00:50:53:14
Speaker 2
My God, I didn't realize.

00:50:53:16 - 00:51:02:14
Speaker 3
Oh yeah. It's been, Yeah. So I want to be mindful about your time. How many more minutes we can go? Ten minutes more. Honest.

00:51:02:14 - 00:51:05:22
Speaker 2
Yeah, definitely. It is so much fun talking to you. So let's.

00:51:05:22 - 00:51:18:16
Speaker 3
Just go. I know it's like, voice, I just, I think, yeah, I know you're doing a podcast with your friends. If I of the proposed to a podcast together, which is talk.

00:51:18:18 - 00:51:40:21
Speaker 3
Man, I really want so here's the thing. I really would like to get to the bottom of sort of the ins and outs or nitty gritty of Nate and the design process. Yeah. The compromises. I hate compromise, but the trade off that you had, you guys had to do to get it off the ground, maybe your take why it became such a sensation right now.

00:51:40:22 - 00:51:42:00
Speaker 2
Yeah. Yeah.

00:51:42:02 - 00:51:57:07
Speaker 3
So that's something that's something I'm interested in because I just don't understand. It's abnormal to me that, like, watching is having a glass of red wine and watching and automation's like, oh, that's great.

00:51:57:09 - 00:52:25:21
Speaker 2
That's fantastic. That's absolutely amazing. So I should preface this by saying that I was there, very late. So, and it was already, I think, three years old by the time I joined. And I was there for about a year working on the AI stuff, scaling up the, design team and, then I went on to other things, but, basically ended and started by John R, co-founder, he co-founder.

00:52:25:21 - 00:52:50:13
Speaker 2
He's he's the founder. John basically started this himself, and he had an epiphany while working with, what's it called? It's like video editing software. Have you seen these, like, node based video editors? Yeah, yeah, he was working with that and he realized, wait a minute. This is very, very similar to how like functional programing works.

00:52:50:14 - 00:53:12:06
Speaker 2
In fact, this is probably functional programing underneath the hood just adapted for video. And he was like we could do this for automation. And instead of writing new programs for everything, we could just encapsulate bits and pieces into nodes. And then you can trigger them when you need them. And he went out and he built a demo very quickly.

00:53:12:06 - 00:53:39:10
Speaker 2
It was open source at the time. And, everyone freaked out because they thought this was absolutely genius. Now, at the time, there were already other no code tools out there, right? There were a lot of things out there that were doing the same thing, doing the exact same thing, but they didn't have this metaphor. So in ended and everything you can do is a node and you hook these nodes up to each other, and then you can literally watch the data flow from node.

00:53:39:10 - 00:53:40:12
Speaker 2
Dutiful.

00:53:40:14 - 00:53:48:06
Speaker 3
Beautiful when you set up the whole thing and then you see that it's moving and it's just the factory lines in the work.

00:53:48:08 - 00:54:15:07
Speaker 2
I care. That's amazing. Right? It's the first time you've really visualized what code is literally doing. Yeah, right. And it's so incredibly amazing to have. And so the, the thing that he thought he was, was doing for people was basically creating this playground where, you only need to connect to external sources once, and then you can reuse that connection in any automation you want to run.

00:54:15:12 - 00:54:33:22
Speaker 2
So for instance, if you want to work with, with Google Sheets, you make a connection once, which is annoying. It's still somewhat annoying in the end, even though it helps you, but then you can just reuse that connection forever. Whereas if you're doing this with code, you have to redo that connection every time you set up a new project.

00:54:33:24 - 00:54:51:15
Speaker 2
So you're never going to do these like tiny throwaway things you're never going to do, like home automation with that because it's super annoying. But then again, it was just incredible. This thing he realized, though, was as soon as he started building that out, people were building all sorts of interesting things, and I done automations for stuff at their home.

00:54:51:15 - 00:55:13:01
Speaker 2
Yeah, but they were also automating complete businesses. And there are entire projects, products built on top of anything, and you can have an entire back end in at a then because he built it open source originally, it was also like start it as something that you can self-hosted, so you don't have to think about the technical aspects of this.

00:55:13:01 - 00:55:26:07
Speaker 2
Now if you want to try it out, you can just go to an NGO and set up an account and get going. But if you have like corporate clients and you want to run automations, you can also just download end and run them on prem.

00:55:26:07 - 00:55:27:16
Speaker 3
But I didn't know that. Wow.

00:55:27:16 - 00:55:50:17
Speaker 2
Yeah. And this was just a side effect of him starting this as a side project because he like, this is what he thought it was going to be a side project that was basically a self-hosted way to automate your stuff. And then they realized, oh wait, we can do so many things with this. So he scaled up a team around him, and they've just been explosively expanding ever since.

00:55:50:22 - 00:56:17:23
Speaker 2
But with AI, particularly something really, really took off. So I came in maybe 3 or 4 months before this AI project happened and I came aboard. We started doing like basic research about a lot of stuff. We were all talking about AI every day. We had a 1 or 2 developers only doing ideation on AI things, and there was all of those like common things, you know, like, how can we how can we add chat?

00:56:17:23 - 00:56:39:14
Speaker 2
Or can we do a chat bot that like, does this for you, blah blah blah. But we quickly found that, like, we can't figure out what people want to do with AI. And the reason for this is the same reason why yawn was confused about what people were using at it. Unfortunate, because when you're building a tool that's as generalizable as automation, editing is even more than automation.

00:56:39:14 - 00:57:03:10
Speaker 2
It just does everything in its code. Right. So now one of the designers there, he's he's pivoted to becoming a developer relations guy. His name is Max, and he's basically he's calling it flow grabbing because it's not programing, because they're not writing code. You can reach the code and write code in anytime if you want to. And you can add like code notes and stuff, but there's no reason for it.

00:57:03:10 - 00:57:22:14
Speaker 2
You can just you can just drag and drop notes and tell you what to do. And it's just it's fascinating to see because it doesn't have like a it doesn't have a clear target market. It has like 13 clear target markets. So when we saw AI tools came out, we looked at all of these basic things like it should be out of chat bot.

00:57:22:14 - 00:57:45:13
Speaker 2
Should we make this help you write your flow whatever. And we we start experimenting with that. But in parallel we John found this that this other open source project called line flow no line chain sorry, line flow is a product and line chain was the same thing. So it's it's a node based way of handling how lambs go through steps.

00:57:45:15 - 00:58:04:05
Speaker 2
So when you're using something like ChatGPT, it's doing a bunch of things manually which have nothing to do with the app. So you give it a prompt, you send it text and it gives you something back. That's the basic thing it does. But as soon as you send it a second message, you're not actually sending it a second message.

00:58:04:05 - 00:58:21:18
Speaker 2
It doesn't have a memory of what happened. What you're doing is you're sending it to the entire history of what you've been talking about, and it reacts to the entire history. If it wants to go online and find something for you, it needs to have a function that it can call. So what it does is it gets a prompt from you.

00:58:21:18 - 00:58:45:22
Speaker 2
It interprets that as, oh, I should call a function online. And then I'm going to get something back and I'm going to present that back, something. Now all of these steps, they presented those as nodes on a graph. And Yan looked at this and was like this slots right in to anything. And so now you could put all of these discrete steps in how you build really cool AI products, right into your graph in any time.

00:58:46:02 - 00:59:10:24
Speaker 2
And you could do them, you could use them for building specifically like agents, or you could use them to like, like push data in and out of other nodes. So you could give an LM memory of something that it didn't have previously, like this stuff that it should know because you put it out, you dragged it out of the database somewhere and shoved it into the into the conversation without the user seeing it and stuff like that.

00:59:10:24 - 00:59:31:17
Speaker 2
So it really, it just meshed really, really well with a metaphor that wireless was going on. We were trying to figure out like, how do we how do we show this? How do we like describe this to the user? And oh my God, we must have made 20, 30 different sketches of like, how that what this could look like and how we could present this.

00:59:31:17 - 00:59:47:03
Speaker 2
And at the end it was just like we just need to ship it, and see how people react. So we took one of the very basic ones that we just shipped that and the weirdest thing is that we still don't know why the design of an end is so useful, but it also.

00:59:47:03 - 00:59:50:02
Speaker 3
Thank you for your honesty. Thank you for.

00:59:50:04 - 01:00:07:20
Speaker 2
It's it's the truth. Like this. This was not my design. I wasn't there for it. This was Max and Yan, figuring this out at the beginning, and we don't know why it's so good. Yeah, it wasn't not it wasn't a a clear vision of where they were going. It was a stepwise process of can we add this?

01:00:07:20 - 01:00:18:18
Speaker 2
Can we add that? And what they've come up with is something that is slightly strange as a solution, but it's really, really simple to learn and super useful.

01:00:18:20 - 01:00:40:13
Speaker 3
As sort of a closing remark, I want to say on the topic, how did, working with Nate and team for considerably long time for like a sort of consultant forward slash entrepreneur, affect change, evolve your design practice or the way you look at design, especially at the intersection of AI.

01:00:40:14 - 01:01:01:01
Speaker 2
So I think this is a tough question, right? Because you're constantly changing. But I think one of the big, big takeaways from Dan was just how powerful it was just to try things, to not overthink them. Oh yeah. Yeah, I know, and this is I think this is something that you have to just learn over and over and over again.

01:01:01:01 - 01:01:21:10
Speaker 2
But without it end, it was so incredibly clear that a lot of times we just thought that this is going to fail and it just does. So we just tried a bunch of stuff that people didn't really believe in, like there were there was no like, strong proponent of it. It was just like, we we have time to do this.

01:01:21:10 - 01:01:46:21
Speaker 2
Let's just try this and see what happens. And it worked. And there were too many of these things to for me to like, disregard as beginner's luck. Eventually I realized that, like, I need to relearn how fast we can do something, and especially with AI, because with AI, like this is a greenfield, like there are no constraints currently or like there are no that there's not a lot of constraints.

01:01:46:23 - 01:02:03:15
Speaker 2
Nobody knows where this is going. And you can just you can just do stuff. You can just do a lot of stuff and it can be weird as hell, right? It doesn't matter. You can just do a lot of stuff and it's it's okay if it fails. Just don't break anything for the user. If they don't like it, it's fine.

01:02:03:15 - 01:02:24:09
Speaker 2
You can just remove it. So there was a there was a ton of super rapid iteration, especially with the AI stuff. We had this, very small team of users who were working with us. They were they were basically given access to a dev branch of any N, and they could use the AI tools, and we used their support as like the feedback for the product.

01:02:24:09 - 01:02:51:19
Speaker 2
So they would ask us like, why doesn't this work? And we were like, we have no idea. Like, what is this? We we don't know, let's figure this out. And we figured it out and we would help them solve it. And we realized like, oh my God, this thing has to like, work differently. And we had this like several times this happened with, with vector databases because, me and one of the PMS, for some reason, we just couldn't figure out what a vector database was at the beginning.

01:02:51:19 - 01:03:11:21
Speaker 2
So one of the developers would come into the room like every other week and be like, guys, you've done it again. Like the vector based vector database doesn't work this way. And it's just like, what did the entire thing like, I think the AI project was like three months from like takeoff to ship, three months.

01:03:11:23 - 01:03:31:23
Speaker 3
That's fast. Yeah. You know, I've got a lot of like, negative feedback. And I wrote on LinkedIn, hey, if I were a senior designer, I would quit my job and I would just explore the AI because exactly. That is, you know, you really need extreme dedication. Like live and breathe, wake up, go to bathroom, take a shower, come back.

01:03:31:23 - 01:03:58:17
Speaker 3
Is thinking about AI coming to your point, it's a wide base. It's just. Yeah, blank canvas is infinite. And I think I, I think you've arguably worked on one of the most groundbreaking products of this era. I think anything's going to be is that not if an end doesn't last or whatever the case that the pattern it's introduced is going to last.

01:03:58:23 - 01:04:19:24
Speaker 3
It's going to go it's. Yeah, obviously. And and that's just very sobering to me. It's just exactly like this. You have to shift if you don't know it, it doesn't about do something about it. You have to get out there in three months. That's how fast things are. Yes. It's it's been a long. I think we need a break.

01:04:20:03 - 01:04:42:13
Speaker 3
We probably can do another part two. But, folks, as promised, after 60 minutes of talking and we talk about NT and everything else. Yes, I'm going to make promise you guys. I'm going to bring him up, bring Jasper back for webinar for another podcast or any other, any excuse. But Jasper, you are you are, you have your own podcast.

01:04:42:13 - 01:04:52:03
Speaker 3
I'm sure you're going to get a lot of followers out of this and you create a mean, maybe do some some plugging and how the audience could find you.

01:04:52:05 - 01:05:13:10
Speaker 2
Oh, with pleasure. So you can find me on. Yes, babylon.com. But if you want to listen to more random design ramblings, I do a podcast. It's, approximately once or twice a month with, an even more senior designer, called you on all of it. And you can find that on design podcast.org. It's called the All Design Podcast.

01:05:13:12 - 01:05:30:07
Speaker 2
And of course, if you need research or if you're doing research, check out find the insights that I, it can help you scale what you're doing really, really fast. And but I just want to say thank you so much for having me on. As usual. This is incredibly fun. And, yeah, we could probably talk for another six hours.

01:05:30:09 - 01:05:37:20
Speaker 3
I think so, I think so, thanks a lot, folks. This was I'm really happy that this happened. Thanks.

01:05:37:22 - 01:05:38:17
Speaker 2
So thank you.

01:05:38:23 - 01:05:50:16
Speaker 1
Thank you for listening to UX for AI. Join us next week for more insightful conversations about the impact of artificial Intelligence in development design, and user experience.