สำเนาบทสนทนา 🔗
WEBVTT
00:00:00.160 --> 00:00:08.720
Machines can make something aesthetically pleasing and they can follow a trend of what's uh you know on brand or something like that.
00:00:08.960 --> 00:00:14.080
But until they've lived a life, they don't have feeling.
00:00:19.280 --> 00:00:23.280
Luke, I've known you for over a decade.
00:00:23.519 --> 00:00:23.839
Okay.
00:00:24.160 --> 00:00:24.559
Really?
00:00:24.640 --> 00:00:25.760
Well, time does fly.
00:00:26.000 --> 00:00:26.960
Yahoo days.
00:00:27.199 --> 00:00:27.359
Yeah.
00:00:27.440 --> 00:00:30.239
Oh, that that may be more than a decade.
00:00:30.559 --> 00:00:31.920
I think it's been more than a decade.
00:00:32.079 --> 00:00:32.240
Yeah.
00:00:32.479 --> 00:00:36.159
And I've watched the journey of you after Yahoo.
00:00:36.880 --> 00:00:45.119
Mobile first, being a prolific designer in terms of what interface shifts we're seeing.
00:00:45.359 --> 00:00:52.479
So what I wanted to focus on today was just this disruption of interface in AI.
00:00:52.719 --> 00:00:54.799
And I mean, it really truly is a disruption.
00:00:54.880 --> 00:00:56.960
It's not just a word that we're using.
00:00:57.200 --> 00:01:03.119
The way that my agency works right now is completely shifts month to month, week to week.
00:01:03.439 --> 00:01:04.799
We're evolving.
00:01:05.200 --> 00:01:11.519
So this conversation is really like to get from you all of the insights of what you're seeing.
00:01:12.000 --> 00:01:19.359
And what is the state of the market, and how do we evolve, one as designers, but also as companies.
00:01:19.680 --> 00:01:25.840
So uh I kind of got started back in the uh mid-90s.
00:01:26.079 --> 00:01:32.799
I was working at the National Center of Supercomputing Applications, and they had a Mosaic, which was the first graphical web browser.
00:01:33.120 --> 00:01:43.519
And so the first things we did for Mosaic is we would be using it for like science sharing, and I built the system for like controlling earthquake shake tables through a web browser.
00:01:43.760 --> 00:01:48.319
Like that was the conception of what the World Wide Web was going to be.
00:01:48.400 --> 00:01:51.120
It's like it's for scientists to share data.
00:01:51.359 --> 00:01:57.680
And so there I lived through that transition from that early concept to what the World Wide Web became, right?
00:01:57.760 --> 00:02:03.359
From there, I ended up going to eBay, building a bunch of new products for them, went over to Yahoo.
00:02:03.439 --> 00:02:14.560
And then similarly at Yahoo, we had all these different products, and I remember a light bulb going off in my head when I saw Flickr translated to a tiny little screen.
00:02:15.199 --> 00:02:22.159
Because what I uh saw with that, I was like, wow, this is so focused on what Flickr is all about.
00:02:22.319 --> 00:02:22.479
Yeah.
00:02:22.639 --> 00:02:30.400
It was like 90% of the interface went away, and it was just the core of, oh, and here's my photos, and I'm looking at photos from my friends.
00:02:30.479 --> 00:02:32.639
There's like three things that you could do with it.
00:02:32.800 --> 00:02:35.039
And that was where the light bulb went off in my head.
00:02:35.120 --> 00:02:41.039
Oh, mobile experiences actually shouldn't just be us cramming, you know, the website in there.
00:02:41.280 --> 00:02:44.879
So that was the basis of mobile first and all that kind of stuff.
00:02:45.039 --> 00:02:48.000
Um, and then from Yahoo, I went to go do startups.
00:02:48.159 --> 00:02:55.360
So did a company called Bag Check that we sold to Twitter, or did another one Polar that Google bought a little bit later.
00:02:55.599 --> 00:03:10.400
And uh it was at Google that I got a bit immersed in a lot of the AI things because internally at Google, like AI was a thing way before any chat GPT, any of this stuff, right?
00:03:10.639 --> 00:03:18.639
And I remember a couple things there that were almost similarly eye-opening as the things that I was describing there.
00:03:18.800 --> 00:03:31.759
Uh we had Google Translate, and all the research scientists and engineers were so excited that the back-end system that they had never taught to translate, I think it was from Korean to Japanese or something like that.
00:03:31.919 --> 00:03:39.759
Like they had never explained to it how to do it, but it learned on its own because it knew how to go from like Japanese to English and Korean to English.
00:03:40.000 --> 00:03:40.159
Okay.
00:03:40.800 --> 00:03:44.400
And I was like, oh, this is like a machine learning.
00:03:44.479 --> 00:03:46.159
That was the whole idea of machine learning.
00:03:46.800 --> 00:03:48.159
And the name came from Luke.
00:03:48.479 --> 00:03:49.520
No, it didn't come from me.
00:03:49.840 --> 00:03:52.800
They had they they called it that for there was actually a running joke.
00:03:52.960 --> 00:03:54.560
There was this video at Google.
00:03:55.360 --> 00:04:00.560
They made like this promotional video of like some kid or you know, college student.
00:04:00.719 --> 00:04:03.599
And in the video, he's like, How was I gonna do it?
00:04:03.759 --> 00:04:04.639
Machine learning.
00:04:04.879 --> 00:04:07.439
So everybody kind of like made fun of that for a while.
00:04:07.599 --> 00:04:11.520
But in the end, right, that was another kind of transformational thing.
00:04:11.840 --> 00:04:24.720
Um, and so kind of seeing all three of those things, and maybe even more, I think, um, has given me a bit of a higher level pattern recognition perspective, right?
00:04:25.279 --> 00:04:29.120
I think I I can see that in the evolution of things that you're posting on LinkedIn too.
00:04:29.199 --> 00:04:32.240
I think one of the first tools you did, I think it was called Benchmark.
00:04:32.560 --> 00:04:33.120
Bench, yeah.
00:04:33.360 --> 00:04:33.680
Bench.
00:04:33.920 --> 00:04:34.160
Yeah, yeah.
00:04:34.319 --> 00:04:35.279
I wasn't one of the first.
00:04:35.360 --> 00:04:37.120
I was like in the middle
Load full transcript
range of whoever it was.
00:04:37.439 --> 00:04:39.360
Okay, but yeah, that was one of them for sure.
00:04:39.680 --> 00:04:45.040
So from Bench, like tell us a little bit about what that was and what has the evolution been since then.
00:04:45.439 --> 00:04:48.319
Well, maybe it's helpful if I take a little bit of a step back.
00:04:48.720 --> 00:05:16.720
So uh one of the first things that we tried to do with like GPT three and a half or three or something like that was we tried to build the system called that we called Enterprise Brain, which was like pretty obvious that you would want to be able to talk to something that understands your organization and what it does and what it's done in the past and what it should do in the future.
00:05:16.959 --> 00:05:28.639
So we started a system where we would just ingest like everything in a company, you know, all the Slack, all the Notion, all the Word docs, uh Google Docs, and those kinds of things.
00:05:28.879 --> 00:05:40.480
And uh it became really hard back then to figure out like what good is because you have all this information and you put it into a system and out comes an answer, and you're like, is that the right answer?
00:05:40.639 --> 00:05:40.800
Yeah.
00:05:40.959 --> 00:05:41.279
I don't know.
00:05:41.360 --> 00:05:43.759
I kind of have to go back and read all these docs.
00:05:43.920 --> 00:05:49.040
Is it if if you remember, you know, met a couple years ago, like hallucinations were still a really big thing.
00:05:49.360 --> 00:05:52.560
Nowadays people rarely talk about it, but still happens.
00:05:52.800 --> 00:06:02.879
Um so we tried to do that, and uh really quickly it became apparent that until we could figure out what a good answer was, we couldn't really build a good system.
00:06:03.439 --> 00:06:10.399
So what I did was I actually repurposed a lot of that technology to my writings over the past like 30-something years.
00:06:10.879 --> 00:06:18.319
So I have 2,000 articles, three books, several hundred like presentations and interviews and all that.
00:06:18.399 --> 00:06:21.759
And I kind of rebuilt the system just using that information.
00:06:22.319 --> 00:06:30.079
And what that allowed me and us to do is to have an eval where I could be like, oh, that's good, because it's my writing, right?
00:06:30.160 --> 00:06:32.879
So technically I'm like Luke, it's of course great.
00:06:33.199 --> 00:06:36.000
Well, no, I can be like, that's not good, this is good.
00:06:36.319 --> 00:06:42.160
And through that, like we learned a lot about how to tune things and adapt and change stuff.
00:06:42.319 --> 00:06:47.439
Um but this comes back to my earlier point about like being in the weeds and actually trying to make things.
00:06:47.600 --> 00:06:59.199
Uh it turned out really early on that being in the weeds uh required like a depth of content understanding that was hard to get when we were working on like these more generic things.
00:07:00.000 --> 00:07:08.399
But I bring up that whole story because from there, like the capabilities of LLMs got better and they could do more things.
00:07:08.560 --> 00:07:13.600
And so then we kind of thought, okay, maybe this is a time to revisit that earlier concept.
00:07:14.160 --> 00:07:24.480
And what we tried to do with Bench was make a set of tools for people who quote unquote do knowledge work that's very similar to that original idea, right?
00:07:24.639 --> 00:07:38.879
Like it would go and research documents for you, it would write you PowerPoint files, it would make you Word docs, it would do spreadsheets for you, a lot of stuff that you now have inside of kind of the core AI tools.
00:07:39.040 --> 00:07:49.439
Uh, but we were in the in the details pretty early on when it was really hard to make these things actually work because the capabilities of the models weren't there yet.
00:07:50.079 --> 00:07:56.720
A kind of a big step forward we had was when Anthropic put out, I think it was Sonnet 3.5.
00:07:57.279 --> 00:08:02.319
Like then that thing would actually allow us to use tools and make things a little bit more agentic.
00:08:02.560 --> 00:08:06.240
But before that, it was like pulling teeth and a bunch of the other models.
00:08:06.720 --> 00:08:20.800
And I think that's another um relevant trend is you kind of build things and you hit a wall, and then up pops a new capability set that comes with a new model, all of a sudden you can do a lot of things that you couldn't do before.
00:08:20.959 --> 00:08:40.000
And that's one of the things I think that makes it hard to keep up and feels like everything's moving so fast because you get these sort of like step function uh capability bumps with these models, and then either the stuff you were doing before becomes irrelevant or it becomes possible, and you're not really sure which one it's gonna be.
00:08:40.320 --> 00:08:40.559
Right?
00:08:40.720 --> 00:08:46.240
Like, did it completely obliterate my startup, or did all of a sudden my startup become viable?
00:08:46.639 --> 00:08:50.480
Um, so that's both unnerving, but also kind of fun, right?
00:08:50.559 --> 00:08:52.240
Because you get all this new stuff.
00:08:52.559 --> 00:08:56.879
It it I have to say, the fun part is what I'm really truly enjoying right now.
00:08:57.039 --> 00:09:14.480
I do feel for the first time completely unlocked recently because of the capabilities of the models, fully unlocked in terms of being able to think of an idea as a creative, take that idea and make it viable and usable.
00:09:14.799 --> 00:09:19.039
And the amount of time I can do that now is so much shorter than before.
00:09:19.279 --> 00:09:24.960
When we worked at Yahoo, I remember I'd have ideas and we would go talk to the engineering team and be like, Can we build this?
00:09:25.120 --> 00:09:26.720
This is such a good idea, we should do this.
00:09:26.799 --> 00:09:28.960
This would help our users, this would help their workflow.
00:09:29.279 --> 00:09:32.720
And the
engineers would be like, No, it's not possible.
00:09:32.960 --> 00:09:38.080
But even in the things that were possible, it would take months, years even.
00:09:39.200 --> 00:09:43.600
And now we're talking about having an idea down to a week, even a day.
00:09:44.240 --> 00:09:49.360
So what have you seen in that shift like in your in the companies that you even work for?
00:09:49.519 --> 00:09:56.000
How is that enabling them, or how how is it increasing the amount they can scale, or it's diminishing?
00:09:56.399 --> 00:10:01.200
Yeah, I I think every it's a double-edged sword, is the way I think about it.
00:10:01.360 --> 00:10:09.679
Um it's very true that you can like whip something up really quickly, but because you can, you don't really have constraint anymore.
00:10:10.159 --> 00:10:13.279
And constraint is valuable, right?
00:10:13.440 --> 00:10:26.000
Like I heard somebody lamenting last week to me about like, you know, it used to be that the time we had to think about things actually helped us call our ideas and get things crisper.
00:10:26.799 --> 00:10:34.000
Then now everyone's just like, oh, I'm gonna build this, I'm gonna build that, I'm gonna build this, and that like refinement process doesn't happen in the early stages.
00:10:34.159 --> 00:10:36.000
I think it still happens a little later.
00:10:36.399 --> 00:10:43.039
Because unless I mean there's lots of companies that are shipping everything and anything, but they really quickly find out that doesn't work very well.
00:10:43.600 --> 00:10:53.519
And then there's a huge difference between making a proof of concept and actually shipping and maintaining and running a product, right?
00:10:53.679 --> 00:10:54.000
Absolutely.
00:10:54.240 --> 00:11:04.159
And uh there's a even though the workload for that has gotten lesser, it's still a lot more than getting that initial idea out, right?
00:11:04.320 --> 00:11:12.960
And so you kind of still have this filter where you have to decide what is worth me putting in that investment time and effort to support.
00:11:13.279 --> 00:11:18.159
I mean, it's not going to be every single vibe-coded idea coming out of a large organization.
00:11:18.320 --> 00:11:21.120
Some of them, yeah, for sure, but not all of them.
00:11:21.440 --> 00:11:22.240
Yeah, not all of them.
00:11:22.720 --> 00:11:30.399
I think, you know, at Babco, we we speak to our portfolio founders in three different phases.
00:11:30.720 --> 00:11:36.720
So we have the starter who's coming in who hasn't gotten money yet from venture, they're not venture-backed.
00:11:36.879 --> 00:11:42.559
Then we have the scaler, those who are, you know, venture-backed, but they're working on scaling their idea.
00:11:42.720 --> 00:11:48.320
And then there's the Titans, those who have already established their ideas, they've already scaled, and now they're just maintaining.
00:11:48.559 --> 00:12:02.879
I'm curious your thoughts on the application of design and LLMs and how because in each
one of those phases, it is true that a starter can get by, for example, prompting and creating quickly and seeing and testing.
00:12:03.039 --> 00:12:13.200
But as you start to scale, there does become that barrier and that filter becomes needs to become really much thicker so that you don't ruin your product for those who are using it.
00:12:13.440 --> 00:12:20.159
But at the same time, there's this disruption that's happening right daily, where people are building their own SAS tools.
00:12:20.320 --> 00:12:20.960
What do they call it?
00:12:21.039 --> 00:12:22.080
I can never say this phrase.
00:12:22.240 --> 00:12:23.519
The SaaS calypse.
00:12:25.679 --> 00:12:26.399
I can't say the same thing.
00:12:26.559 --> 00:12:27.440
I think that's what they call it.
00:12:27.519 --> 00:12:27.759
Yeah.
00:12:28.159 --> 00:12:29.039
The death of software.
00:12:29.519 --> 00:12:32.240
The death of software is easier to say, yeah.
00:12:32.399 --> 00:12:34.000
Um, what are your thoughts on that?
00:12:34.320 --> 00:12:35.360
On the death of software?
00:12:35.519 --> 00:12:37.600
Uh, many thoughts.
00:12:37.759 --> 00:12:41.600
Uh if you look at a lot of SaaS companies, they are bloated.
00:12:42.080 --> 00:12:44.080
And they are not great, right?
00:12:44.159 --> 00:12:51.759
Like your average enterprise piece of software is very hard to use, very hard to maintain, very expensive to maintain.
00:12:52.000 --> 00:12:56.000
So is there an opportunity to like get rid of the crappy parts of that?
00:12:56.240 --> 00:12:57.120
Yeah, absolutely.
00:12:57.279 --> 00:13:02.559
And maybe this kind of drives the reckoning of that, which I think would be good for everybody.
00:13:02.720 --> 00:13:04.879
Uh, do I believe software is going away?
00:13:05.120 --> 00:13:05.440
No.
00:13:05.600 --> 00:13:14.480
Uh we're still gonna, even if you know, you are as AGI pilled as some people are, and you think, you know, that you still are believing in software.
00:13:14.559 --> 00:13:14.720
Yeah.
00:13:15.120 --> 00:13:21.360
You still think that people are gonna be using computers and software is gonna be written by machines and what have you, right?
00:13:21.519 --> 00:13:23.039
There still will be software.
00:13:23.279 --> 00:13:47.279
So the concept of the end of software is a little weird, but it may very much allow some of these new entrants to come in that have been what's happened in previous transitions, right, is you've got companies that were built in a previous era where every single layer is kind of built on the fundamentals of what was going on then, right?
00:13:47.360 --> 00:13:49.440
If you look like a company like Yahoo, right?
00:13:49.600 --> 00:13:55.679
Yahoo grew up on the web and then sort of made it to Web 2.0, but it had a real hard time with mobile.
00:13:56.000 --> 00:14:01.840
Because it wasn't in their DNA, it wasn't in their systems, it wasn't in their design understanding and everything.
00:14:02.080 --> 00:14:10.159
Um, so similarly now,
you know, you've kind of been built up in the age of cloud software, and AI comes up and it affects every part of the layer.
00:14:10.480 --> 00:14:20.240
A company that naturally gets built that way, or uh from its core, gets built that way, if AI has different advantages than that incumbent.
00:14:20.320 --> 00:14:24.799
And so I'm I'm pretty confident we'll get a couple companies, well, we are already seeing it, right?
00:14:24.879 --> 00:14:32.720
But we'll get a couple companies that are very AI native throughout the stack, disrupting some of those software companies.
00:14:32.879 --> 00:14:39.759
Um, but even those software companies, if they transition and make it through, I think they'll look pretty different on the other side.
00:14:39.919 --> 00:14:46.399
All right, maybe smaller, uh, but definitely more embedded with AI throughout the stack.
00:14:46.879 --> 00:14:47.360
For sure.
00:14:47.679 --> 00:14:52.320
And what would you tell those companies on how to work through this disruption?
00:14:52.720 --> 00:14:54.240
Yeah, same thing we were talking about before.
00:14:54.480 --> 00:14:56.720
Like, do it, like start using the stuff.
00:14:57.200 --> 00:15:04.639
I uh I get pinged, I don't know, like once every week or every other week by some company that's like, yeah, we're really late to this.
00:15:04.879 --> 00:15:08.320
Can you come and talk to us about what's going on?
00:15:08.559 --> 00:15:17.840
And my response is always, I can, but you're best off like actually using the stuff and getting in there and figuring out what works for you, what doesn't work for you.
00:15:18.000 --> 00:15:29.600
There's a it there was a group of, I don't know if it's a group or like a mindset, but there was always this kind of, oh, this is hype, it will pass, I don't have to worry about this mindset.
00:15:29.679 --> 00:15:29.840
Yeah.
00:15:30.000 --> 00:15:35.039
When that happened with mobile, happened with web, happened with PCs, happens all the time, right?
00:15:35.279 --> 00:15:40.720
I remember in the early mobile days, I would have similar conversations with companies, right?
00:15:40.799 --> 00:15:45.440
I'd get a phone call from Netflix, hey, can you come in and talk about mobile?
00:15:45.679 --> 00:15:52.000
And I would present all this stuff to them, and somebody would raise their hand and be like, I don't think anybody's gonna watch videos on mobile.
00:15:52.799 --> 00:15:53.759
Oh boy, where they wrong.
00:15:54.080 --> 00:15:57.919
Well, that's I guess that's that's a consideration, right?
00:15:58.000 --> 00:16:00.399
Because they're like, oh, the screen is small, right?
00:16:00.480 --> 00:16:03.440
The bandwidth crappy people watch on big screens.
00:16:03.679 --> 00:16:04.639
That's how they watch.
00:16:05.120 --> 00:16:10.480
Uh similarly, you know, eBay's like, I don't think anybody's actually gonna buy anything really expensive on mobile.
00:16:10.639 --> 00:16:14.879
They might like check their bids and see how it's going, but they're never gonna really like full.
00:16:15.200 --> 00:16:31.600
So there's always there's always the same kind of, I don't know if it's denial, um, but you like feel like it's gonna pass, or you feel like it's not gonna uh grow in the way that it becomes meaningful to you.
00:16:31.919 --> 00:16:35.279
And similarly with AI, like a lot of people had that mindset, right?
00:16:35.440 --> 00:16:46.080
Oh, it's drawing six fingers, so it'll never make good images, or it hallucinates all the time, we'll never be able to use it for real work and you know, add whatever to your list of excuses.
00:16:46.720 --> 00:16:51.840
Uh but over and over again, with technology, like, yeah, it draws six fingers.
00:16:51.919 --> 00:16:54.960
There's like thousands of people working on getting it down the five fingers.
00:16:55.039 --> 00:16:57.039
I think it'll get down the five fingers, right?
00:16:57.360 --> 00:16:58.159
I think we've gotten there.
00:16:58.480 --> 00:16:59.919
Yeah, we've absolutely gotten there.
00:17:00.000 --> 00:17:00.320
100%.
00:17:00.639 --> 00:17:03.039
Same with the we're just talking about the hallucinations, right?
00:17:03.120 --> 00:17:03.759
It was a problem.
00:17:03.919 --> 00:17:05.359
Everybody knew it was a problem.
00:17:05.519 --> 00:17:10.960
Tons of smart people worked on it in lots of different ways, and you don't really hear people bringing it up that much anymore.
00:17:11.359 --> 00:17:12.000
That's so true.
00:17:12.240 --> 00:17:17.839
So use it and and instill it into like your company.
00:17:18.000 --> 00:17:19.759
But the biggest thing to do is to use it.
00:17:20.000 --> 00:17:20.160
Yeah.
00:17:20.720 --> 00:17:20.960
100%.
00:17:21.279 --> 00:17:27.759
You don't know what it's good or bad for, and you don't know what it can do and what it can't do until you're actually trying to apply it.
00:17:28.000 --> 00:17:36.000
It's not just use it, because some people will use it in a way that is like almost like false positive or false negative.
00:17:36.240 --> 00:17:36.400
Right?
00:17:36.559 --> 00:17:40.240
Oh, I did it for this thing that I have no idea about, and it's amazing.
00:17:40.559 --> 00:17:46.400
Like, okay, why don't you try it on a thing that you actually know a lot about, and then see where you're at, right?
00:17:46.880 --> 00:17:55.759
In using it, I truly feel like if you're using it in a way that's impactful to yourself and to your company, it is a game changer.
00:17:55.920 --> 00:17:56.720
It truly is.
00:17:56.880 --> 00:18:02.079
And your brain starts to unlock different, I call them like levels.
00:18:02.400 --> 00:18:10.319
Like if you're playing a video game, like I'm operating at level one without AI, and then level 10 with AI, and then it unlocks something else.
00:18:10.400 --> 00:18:13.839
I'm like, now I'm at level 12, 13, 14, oh my god.
00:18:14.079 --> 00:18:21.839
Like I feel powerful and multiplied, but at the same time, in that feeling, you're like, oh, like
where is this leading?
00:18:22.319 --> 00:18:22.960
Where's the line?
00:18:23.200 --> 00:18:24.160
Where's the line?
00:18:24.640 --> 00:18:28.000
And and I think that the line, like, let's talk about the line.
00:18:28.079 --> 00:18:30.799
Where what are you what do you think the line is?
00:18:30.960 --> 00:18:32.319
And is it a real line?
00:18:32.400 --> 00:18:33.519
Is it artificial line?
00:18:33.599 --> 00:18:36.400
Is it established line that someone needs to say this is the line?
00:18:36.559 --> 00:18:37.440
What does that look like?
00:18:37.759 --> 00:18:46.160
I mean, there's a lot of conversations around what is the role of the human once the computer becomes quote unquote smart enough.
00:18:46.559 --> 00:18:46.720
Right?
00:18:46.880 --> 00:18:49.599
I don't know what the definition of smart enough is.
00:18:50.240 --> 00:19:02.720
But uh I feel over all these years, every time we've made tools, right, we have had people who use those tools to achieve stuff.
00:19:02.960 --> 00:19:03.200
Yeah.
00:19:03.359 --> 00:19:06.079
Uh so I don't think it's gonna be that different this time.
00:19:06.160 --> 00:19:09.519
And this is again while why using the tools makes a ton of sense.
00:19:09.759 --> 00:19:13.519
Will those tools completely replace certain things that humans used to do?
00:19:13.680 --> 00:19:14.160
Sure.
00:19:14.480 --> 00:19:14.799
Right?
00:19:14.880 --> 00:19:20.000
Like we have so many examples of that throughout the world, and jobs change as a result of it.
00:19:20.160 --> 00:19:24.880
Nobody's like uh building wheels for horse-drawn buggies anymore.
00:19:25.039 --> 00:19:27.920
Maybe there's like one guy somewhere, right, or two or three.
00:19:28.000 --> 00:19:28.240
I don't know.
00:19:28.559 --> 00:19:30.079
It's probably pretty popular for people who want that.
00:19:30.559 --> 00:19:30.799
Sure.
00:19:30.880 --> 00:19:36.160
Yeah, like there's a couple of people, but the majority of the world is not doing that anymore.
00:19:36.240 --> 00:19:39.039
But before it used to be, you know, like a really big occupation, right?
00:19:39.200 --> 00:19:41.200
As just one micro example.
00:19:41.279 --> 00:19:43.119
There's tons and tons of those.
00:19:43.279 --> 00:19:45.440
And so I that's gonna happen, right?
00:19:45.519 --> 00:19:50.480
The kinds of stuff that you had to do before you had these capabilities uh won't exist.
00:19:50.640 --> 00:20:00.240
The open question, I think, is like how good will these systems get, and how far can their autonomy go?
00:20:00.640 --> 00:20:03.119
And to that I don't have an answer, right?
00:20:03.519 --> 00:20:19.680
My uh maybe kind of optimistic view is I think they will get to the level of like a very smart human, which is perhaps as far as they can go, maybe they can go to like superhuman.
00:20:19.839 --> 00:20:23.279
But even today, if we have a bunch of smart humans, we call them a team.
00:20:23.599 --> 00:20:26.960
And they build a company and they work together,
and right?
00:20:27.359 --> 00:20:32.960
So, like we have a model of what if we if we have something that's really a great coder, how do we use that?
00:20:33.039 --> 00:20:34.240
We know how to use that.
00:20:34.319 --> 00:20:36.960
Or if we have something that's like a really great designer.
00:20:37.279 --> 00:20:41.039
You've worked on many teams like that, I've worked on many teams like that.
00:20:41.279 --> 00:20:50.480
So it's not that strange to think, oh, here's you know, like a designer I can spin up and down, here's a developer that I can spin up and down.
00:20:51.039 --> 00:20:57.200
Um, I know how to you know work with that and direct and guide it and those kinds of things and fit it in.
00:20:57.519 --> 00:21:07.839
Um what I will say is I think so far most of these AI tools have been very much like solo sports.
00:21:08.240 --> 00:21:13.359
And this is why you hear people being like, oh, I don't even need developers anymore, I can just type it in the box.
00:21:13.599 --> 00:21:17.680
Or developers, I don't even need designers anymore, I just type it in the box.
00:21:19.039 --> 00:21:20.000
Whatever you want to say.
00:21:20.400 --> 00:21:23.920
Because these things have been built like individual empowerment.
00:21:24.640 --> 00:21:29.680
What we haven't gotten to yet is how do these things fit into like collaboration.
00:21:29.839 --> 00:21:30.000
Yeah.
00:21:30.240 --> 00:21:44.000
So we've been working a bunch on those types of projects where like a developer and all the things they know and understand, they can lean deeply into that, and the designer, all the things they know and understand, can lean deeply into that.
00:21:44.400 --> 00:21:48.000
But they can collaborate in a way that brings both the worlds together.
00:21:48.160 --> 00:21:48.319
Yeah.
00:21:48.480 --> 00:21:55.119
So you don't have to, as a designer, all of a sudden become a developer, and as a developer, you don't have to become a designer, right?
00:21:55.279 --> 00:21:59.680
Um, and I think that's probably gonna be a big area of focus.
00:22:00.079 --> 00:22:00.160
Yeah.
00:22:00.559 --> 00:22:04.559
Going forward, because what you don't want is what we currently have.
00:22:04.880 --> 00:22:07.200
Okay, I'll take another step back here.
00:22:07.599 --> 00:22:26.880
Uh, ever since the job roles of user experience professionals and designers, and the job roles of things like front-end engineers and software developers became not only a thing, but a thing a lot of people wanted to do, there's been uh markets that kind of cater to those people.
00:22:27.279 --> 00:22:32.000
Here's your new design tool, here's your new development framework, here's your deployment system.
00:22:32.079 --> 00:22:40.160
And so these groups have built up these like tool stacks which are very specific to them, right?
00:22:40.480 --> 00:22:44.799
And so we've actually made the process of throwing things over the wall worse.
00:22:45.519 --> 00:22:49.759
Because like, go in and try and figure out AWS as a designer, good luck.
00:22:50.000 --> 00:22:55.200
Go in and try and learn Figma auto layout as a developer, God bless you, right?
00:22:55.759 --> 00:22:58.079
The walls just have been getting higher and higher.
00:22:58.400 --> 00:23:02.559
And we've been throwing things over the wall for many years now because of that.
00:23:02.799 --> 00:23:06.799
And these current AI tools, they just sort of like add more things to the list.
00:23:07.359 --> 00:23:11.119
And you can make things more, you can make more things and you can make them faster.
00:23:11.200 --> 00:23:12.799
So you're throwing more stuff over the wall.
00:23:12.960 --> 00:23:13.359
Yeah, yeah.
00:23:13.680 --> 00:23:13.839
Right?
00:23:14.079 --> 00:23:16.240
As a designer, you're like, oh look, I use claude code, here you go.
00:23:16.640 --> 00:23:22.640
Boom, boom, boom, boom, and all these things are f and then the developer's like, I made this prototype, and the PMs, I made this, and you're like, whoa, whoa, whoa.
00:23:23.039 --> 00:23:23.519
Right?
00:23:23.680 --> 00:23:26.319
Um, not very good for collaboration.
00:23:26.480 --> 00:23:26.640
Yeah.
00:23:26.960 --> 00:23:30.400
Collaboration works better when you all work on the same thing, right?
00:23:30.480 --> 00:23:34.079
And you can all bring the things that you do well to bear on it.
00:23:34.240 --> 00:23:41.519
So that's pretty missing right now, but pretty obvious that I think something like that's gonna have to happen and change.
00:23:41.839 --> 00:23:45.519
Because again, we work on teams, we work with other people.
00:23:45.920 --> 00:23:52.079
And I do think there's something to be said, you know, about a role and being an expert in that role.
00:23:52.400 --> 00:24:08.880
The designer's role, an expert coming from the user perspective, understanding, you know, spacing and pixels and being trained in communication and flow and journey, and you know, each person being able to have that space and direction over the interface.
00:24:09.119 --> 00:24:16.559
And when you say it's very solo right now versus team, again, in the starter world, you can get away with being solo.
00:24:16.799 --> 00:24:18.480
You can I have an idea, I'm gonna go do it.
00:24:18.559 --> 00:24:23.200
And there might be just one designer and one engineer, and they're working on it, and they're working on it, and they're working on it.
00:24:23.359 --> 00:24:31.359
When you start to scale something, and there's an interface that's maintainable, that needs to be maintained, that has users on it, and you need to add things to it.
00:24:31.599 --> 00:24:35.519
And then you talk about all the way up into a Titan like an Airbnb.
00:24:36.000 --> 00:24:39.759
You can't just then go off and go code something up and then put it in.
00:24:39.920 --> 00:24:41.200
It doesn't work that way.
00:24:41.440 --> 00:24:46.240
You have to be, you have to plan it strategically, and you have to have that oversight and insight.
00:24:46.480 --> 00:24:48.400
Um, so that's what we've been doing.
00:24:48.640 --> 00:24:52.880
Yeah, something has to kind of keep things cohesive and unified.
00:24:53.119 --> 00:24:58.319
Um, I do think actually an AI system can play that role, right?
00:24:58.480 --> 00:25:02.960
If it's shaped by the designer, if it's shaped by the developer and shaped by everybody.
00:25:03.119 --> 00:25:17.039
Like this is one of the things that we've been building is we have a system where the designer can kind of bake in the design system, the developer can bake in the frameworks and the structure and the deployment and all of that.
00:25:17.200 --> 00:25:26.880
And then once that exists, anybody can kind of come in and do something that what pops out on the other side is more unified or an alignment to the system.
00:25:27.119 --> 00:25:27.279
Yeah.
00:25:27.920 --> 00:25:33.440
Follows the dev rules, follows the design rules, but you can go in there and like make this and like boop.
00:25:33.519 --> 00:25:33.839
Right?
00:25:33.920 --> 00:25:36.480
So you're not living in all these like separate worlds.
00:25:37.440 --> 00:25:46.559
Um and I think that becomes much more a way to apply this to the scaling world that you're talking about, or the Titan world that you're talking about, right?
00:25:46.880 --> 00:25:47.440
Completely.
00:25:47.599 --> 00:25:51.279
And so what do you imagine is the new team?
00:25:51.519 --> 00:25:55.839
Like, how do they work differently than how we used to work at Yahoo?
00:25:56.079 --> 00:26:01.839
What is because I feel like for a long time, the way we worked at Yahoo was how people worked, how teams worked.
00:26:02.240 --> 00:26:04.480
And that continued for a while.
00:26:04.720 --> 00:26:06.799
So, like, what's that shift look like now?
00:26:07.200 --> 00:26:10.079
Yeah, so let's talk about some of those things.
00:26:10.240 --> 00:26:12.640
One I think is the collaboration piece.
00:26:12.880 --> 00:26:21.440
The other one that I think is um for years, people's jobs have been making digital detritus.
00:26:21.599 --> 00:26:23.119
Oh, I'm the user researcher.
00:26:23.200 --> 00:26:27.839
Why hasn't anybody read my hundred page PowerPoint about users?
00:26:28.880 --> 00:26:31.440
Uh and then I'm the, you know, I'm the designer.
00:26:31.519 --> 00:26:37.039
Why hasn't anybody read the 15 pages of guidelines that I wrote about how to implement the design system?
00:26:37.680 --> 00:26:47.200
Or the, you know, like we just keep making all this digital detritus, like filling notions, filling Google Drives, filling PowerPoints, what what have you.
00:26:47.359 --> 00:26:56.880
Um and one of the things that I think I'm optimistic about that AI will do is it'll basically make clear that those things are not very valuable.
00:26:57.039 --> 00:27:00.319
Because I can make them in half, like you want a PDF, here you go.
00:27:00.720 --> 00:27:02.559
You want a 20-page design doc?
00:27:02.720 --> 00:27:03.039
There you go.
00:27:03.119 --> 00:27:06.400
I like literally press a button, type in the box, and it pops out.
00:27:06.640 --> 00:27:14.400
So once that mountain of those kinds of things starts to scale, people will really quickly realize, okay, that's not the work.
00:27:14.799 --> 00:27:15.119
Right?
00:27:15.440 --> 00:27:20.960
Me making a big doc and forcing everybody to read it is not how we should collaborate.
00:27:21.119 --> 00:27:23.279
But it has been how we've been doing it for a long time.
00:27:23.680 --> 00:27:35.680
Come back to the collaboration one, uh, right, throwing things over the wall, been doing that forever, which is one of the things that takes so long in these processes, is I have to translate what you wrote into my thing.
00:27:35.759 --> 00:27:39.920
Like the poor developer has to go in the Figma, look at the screen things.
00:27:40.240 --> 00:27:42.559
Yeah, look at the specs, and then like translate them over.
00:27:42.880 --> 00:27:46.559
And if something changes, then they have to go back and right.
00:27:47.200 --> 00:27:50.799
So like the handoff is like a brutal thing.
00:27:51.039 --> 00:28:03.359
Um so where I think we're at right now is you can basically generate things in real time as they're relevant for what they're relevant.
00:28:03.839 --> 00:28:07.920
Because before every one of these documents that you publish, it's out of date.
00:28:08.480 --> 00:28:11.519
The moment it's published, it's out of date.
00:28:11.599 --> 00:28:18.160
And companies are just filled with this detritus, and nobody's lit knows where where's the source of truth, what is that?
00:28:18.400 --> 00:28:19.440
Did we do this before?
00:28:19.599 --> 00:28:22.079
Why do we don't know?
00:28:22.400 --> 00:28:28.480
But uh to go back to my little uh ask Luke thing that I made of all my writings and things.
00:28:28.640 --> 00:28:32.480
What happens now, if somebody asks a question like, what's an AI agent?
00:28:32.559 --> 00:28:39.039
It will use all that information and create a new thing that is relevant for that person in that moment.
00:28:39.599 --> 00:28:40.000
All right?
00:28:40.160 --> 00:28:44.079
And then after that, gone, useless, right?
00:28:44.400 --> 00:28:58.559
I think that's not only the impact to the workflow, but the impact to the end user that I'm most excited about of being able to open an interface and it be exactly what I need in the moment I need it.
00:28:58.799 --> 00:29:00.559
Before we had feature bloat, right?
00:29:00.720 --> 00:29:06.240
You had 10,000 features in an attempt to make sure that everyone was accommodated for.
00:29:06.799 --> 00:29:12.640
But now, if we use it right, you can open an interface and it can say, these are the three things you need, Olivia.
00:29:13.200 --> 00:29:14.400
You don't need anything else.
00:29:14.559 --> 00:29:22.799
And that's what I'm so excited about, because you I right now I can't tell you how many times I go into an interface and I'm like, you have no idea what I
need right now.
00:29:22.960 --> 00:29:23.839
And it's showing.
00:29:24.000 --> 00:29:24.160
Yeah.
00:29:24.400 --> 00:29:26.640
And now we can, it's like we're not we're not talking about it.
00:29:26.880 --> 00:29:28.720
I don't know if this is I don't know if this is gonna get solved.
00:29:28.960 --> 00:29:29.119
Okay.
00:29:29.440 --> 00:29:31.680
Okay, well, let me give you two thoughts on that.
00:29:31.839 --> 00:29:40.480
One, even if you look at the interfaces of like OpenAI anthropic, which are technically right, like currently bleeding edge, they just keep adding more crap to their interface.
00:29:41.200 --> 00:29:42.480
And now you got this tab and these buttons.
00:29:42.880 --> 00:29:44.480
They move my projects, I can't find them.
00:29:44.880 --> 00:29:45.759
It's all over the place.
00:29:45.920 --> 00:29:47.759
So that problem still exists.
00:29:48.000 --> 00:29:55.519
But in order to actually achieve what you're describing, I think you need infinite context, which I don't think is possible.
00:29:55.680 --> 00:30:00.079
And this is the example I give people of the people on this planet who know me the best.
00:30:00.240 --> 00:30:04.079
Probably my wife, maybe my parents, maybe my kids are pretty high up there.
00:30:04.160 --> 00:30:06.319
But let's just stick with my wife, right?
00:30:07.279 --> 00:30:11.359
See her every day, talk to her all the time, married 20 years, whatever.
00:30:11.599 --> 00:30:14.160
She has no idea what we're talking about here because she's not here.
00:30:14.640 --> 00:30:25.279
She has no idea what I did this morning, what I listened to on my car ride to work, what I saw on my ride to work, like missing so, you know, what's going on in my brain, missing so many other things.
00:30:25.359 --> 00:30:33.119
But all those things at any point in time can actually influence what I need and how I, you know, how I need a tool to work.
00:30:33.599 --> 00:30:38.079
And I just don't know how we build a system that collects all of that information.
00:30:38.319 --> 00:30:44.720
Even if you're like audio video recording everything from birth and you have brain waves in there, right?
00:30:44.960 --> 00:30:45.680
We could we could get there.
00:30:46.400 --> 00:31:00.559
Maybe we could get there, but even then, right, like you would have to not only record everything I'm seeing and hearing, but you would also have to capture my brain waves to sort of know, or whatever, some brain thing, to know like what I think about that.
00:31:00.960 --> 00:31:02.559
What did I relate it to?
00:31:02.720 --> 00:31:09.759
How did I I think we could get there, maybe it doesn't need to be that large of an expansive idea.
00:31:09.839 --> 00:31:11.279
It doesn't need to map your whole brain.
00:31:11.519 --> 00:31:11.680
Yeah.
00:31:12.240 --> 00:31:19.200
But as an idea, like um, my brother-in-law is working on, we're we're all into tracking our macros.
00:31:19.759 --> 00:31:23.119
Um so meal planning and macro planning.
00:31:23.519 --> 00:31:26.480
Every time you open the interface, it's like the same thing.
00:31:26.720 --> 00:31:29.119
The same interface for you, for me.
00:31:29.759 --> 00:31:36.799
But if I am in a certain journey in my macro plan, and I'm like an elite macro planner because like I'm an NPC athlete.
00:31:36.880 --> 00:31:42.640
So my macro planning, what I needed from an interface, would be very different from someone who's just trying to lose a little bit of weight.
00:31:43.359 --> 00:31:48.720
So those interfaces can now be one, it can be fundamentally different for both of us.
00:31:48.880 --> 00:31:56.640
But also, depending on the time of day I open the app or where I'm located, it can start to Yeah, it's just it's it's a slippery slope?
00:31:56.960 --> 00:31:57.599
Slippery slope.
00:31:57.839 --> 00:31:57.920
Yeah.
00:31:58.160 --> 00:31:58.480
Yeah.
00:31:59.680 --> 00:32:01.680
So for some things, for sure, right?
00:32:02.079 --> 00:32:05.920
But again, like we we've built these types of things.
00:32:06.000 --> 00:32:09.279
Like when I was at Google, we built a couple of these things, like Google Now, right?
00:32:09.599 --> 00:32:16.880
Google Now was a stream of cards that was supposed to use all the machine learning intelligence we had back then to like predict what you need.
00:32:17.200 --> 00:32:17.359
Right?
00:32:17.440 --> 00:32:21.200
Oh, it's like lunch, you're at work, you probably need to go to a restaurant.
00:32:21.519 --> 00:32:24.079
Like, no, no, no, I gotta get my butt to a meeting, right?
00:32:24.240 --> 00:32:24.559
Or whatever.
00:32:24.960 --> 00:32:26.000
Like we're skipping lunch today.
00:32:26.319 --> 00:32:26.400
Right?
00:32:26.480 --> 00:32:34.000
Or I brought my own lunch, or any number of reasons why that thing is irrelevant, because it doesn't have the full context.
00:32:34.480 --> 00:32:39.200
Um and so like all the examples there is like, oh, here's your boarding pass for your flight, right?
00:32:39.519 --> 00:32:40.400
It's true.
00:32:40.799 --> 00:32:54.079
Um and I would love to believe in that type of vision, but I feel like unless it's r always right, you're gonna lose faith in it really quickly once it's wrong a few times.
00:32:54.240 --> 00:32:56.480
And that's what happened with a bunch of these products that we made.
00:32:56.559 --> 00:33:01.440
It's like once you're wrong, like once, twice, maybe that's all it takes.
00:33:01.920 --> 00:33:03.680
Then people are like, this thing's stupid.
00:33:03.920 --> 00:33:05.920
Doesn't know what it doesn't know what I need.
00:33:06.079 --> 00:33:07.519
It's showing me this, you know.
00:33:08.079 --> 00:33:18.319
Is it better then to give this do you still think that it's better then to give the same interface to all users still versus trying to adapt to what the user might need?
00:33:18.799 --> 00:33:36.400
I wonder if it becomes a little bit more like there's specialized tools that because they're so fast and easy to make
now, right, that you just have a different tool based on your particular needs rather than falling into this like very big generic type of tool.
00:33:36.640 --> 00:33:36.799
Yeah.
00:33:37.039 --> 00:33:44.079
Um but I don't know, people still using Google Docs today, people still using email today, people still using Slack today.
00:33:44.319 --> 00:33:49.440
Are all those things gonna go away and be subsumed by a single box?
00:33:50.720 --> 00:33:51.440
I don't know.
00:33:51.680 --> 00:33:52.559
So it's a good question.
00:33:52.880 --> 00:33:54.799
There maybe can be levels, right?
00:33:54.960 --> 00:34:11.280
Like if there's an entry level to context, not that you need to know everything about me, but if we look at an interface, the the the the things that I need from the interface, maybe like a font size change, um, or if I I can't hear, I can't see, or I'm color colorblind, right?
00:34:11.440 --> 00:34:14.639
Those are fundamentals of how I operate within the interface.
00:34:14.960 --> 00:34:20.000
Or like maybe I have a need where I get too sidetracked with information.
00:34:20.079 --> 00:34:22.960
So with the amount of features that I even use, right?
00:34:23.039 --> 00:34:28.239
Like this user only uses five percent of the features, and the other 95 are removed.
00:34:28.400 --> 00:34:30.719
Like you can start to Yeah, I love the vision.
00:34:30.880 --> 00:34:32.800
I've the vision makes a lot of sense.
00:34:32.880 --> 00:34:35.679
And for a couple fundamental things, yeah, sure.
00:34:35.920 --> 00:34:38.400
But uh context is very hard.
00:34:38.480 --> 00:34:40.719
So let me say two things on that.
00:34:40.880 --> 00:34:48.079
That system that I talked about earlier, this enterprise brain, right, it would slurp up all the information on all the things happening.
00:34:48.239 --> 00:34:54.000
And granted, the models were kind of a little dumber back then, but if you went and asked it, who is Luke?
00:34:54.239 --> 00:34:57.360
It would say, Luke is a guy who talks on Slack a lot.
00:34:57.599 --> 00:34:58.880
And it was right, right?
00:34:58.960 --> 00:35:06.400
Like if you look at all the documents and see everything, the place where I probably show up the most is like Slack message, slack message, slack message.
00:35:06.480 --> 00:35:11.440
And then there's some little bio of me somewhere buried in like some HR system.
00:35:11.920 --> 00:35:12.159
Right?
00:35:12.320 --> 00:35:15.199
But that's not what I was asking at all when I say who is Luke.
00:35:15.280 --> 00:35:19.679
And maybe this, you know, the system gets smarter, it can kind of figure out that intent, right?
00:35:19.760 --> 00:35:32.079
But there's like a bazillion cases like that where you have this expectation of the person and then the reality of the data to contend with, which is which is hard to do.
00:35:32.159 --> 00:35:41.280
And if if you look at all the adaptive interfaces that we've made over the years, right, I think none of them have really been successful.
00:35:41.760 --> 00:35:48.559
If you you remember Microsoft Office's big redesign where they like remove menus and like it was an endless source of frustration.
00:35:48.719 --> 00:35:50.800
Because you're like, wasn't that feature there?
00:35:50.960 --> 00:35:53.280
No, you didn't use it in three days, so now it's gone.
00:35:53.519 --> 00:35:55.280
You're like, Bring it back.
00:35:55.679 --> 00:35:56.239
What the F.
00:35:56.480 --> 00:36:02.320
But an area that I think is more promising in that area is kind of what we call just-in-time interfaces.
00:36:02.880 --> 00:36:16.400
So we launched some of this stuff in our um creative tool company, Rev, where you can describe the type of filter you want on an image and it'll automatically make it for you and save it for you and keep it for you.
00:36:16.800 --> 00:36:17.039
Right?
00:36:17.119 --> 00:36:25.679
So, but that's not like automatically we decide you want this filter because of where you are and who you are and what you've done before.
00:36:25.920 --> 00:36:29.119
You explicitly say, I want to have a filter that does this and this.
00:36:29.199 --> 00:36:29.599
Okay, cool.
00:36:29.679 --> 00:36:31.199
We'll make that interface for you.
00:36:31.280 --> 00:36:32.239
We'll put it over here.
00:36:32.320 --> 00:36:34.960
If you ever want to reuse it again, it's there for you.
00:36:35.360 --> 00:36:39.440
Or you go in and say, hey, I want everything to be larger font.
00:36:39.679 --> 00:36:40.480
Okay, cool.
00:36:40.719 --> 00:36:43.840
You I didn't infer, oh, this person's eyes look shady.
00:36:44.079 --> 00:36:46.159
Maybe I need to make everything bigger, right?
00:36:47.199 --> 00:36:50.079
Or they were at the eye doctor, let's make everything huge.
00:36:52.000 --> 00:36:54.719
I don't want people to be tracking their whole lives for sure.
00:36:54.880 --> 00:37:02.559
I do think there's opportunity there, though, yeah to be adaptive, as you're saying, in a specific way that they re- that is requested.
00:37:02.880 --> 00:37:04.559
Yeah, that's a manual.
00:37:04.800 --> 00:37:06.559
The requested part is very Right.
00:37:06.719 --> 00:37:12.639
If the if I'm doing it, then by all means, like malleable and adaptable software makes sense.
00:37:12.800 --> 00:37:22.639
It's the part where we're trying to use all this context to automatically change things for you, where at least to date, right, like the track record's pretty bad.
00:37:23.199 --> 00:37:24.239
We haven't done that well.
00:37:24.880 --> 00:37:26.079
So much to solve.
00:37:26.239 --> 00:37:32.559
And I think that there's, you know, some people, there's the there is this fear of like AI coming in and taking over.
00:37:32.880 --> 00:37:35.920
But I feel like there's so much to solve.
00:37:36.000 --> 00:37:41.440
And in all technology that's ever been in existence, it multiplies what we need to do.
00:37:41.599 --> 00:37:43.039
And that is the truth.
00:37:43.280 --> 00:37:53.519
Like, right now,
even the ideas that we're talking about, there's room for opportunity for companies to go explore these ideas and and make their products different and evaluate what's needed.
00:37:53.679 --> 00:37:57.920
And then the workflow shifts, and then there's more work to be done for users.
00:37:58.079 --> 00:38:02.159
And I think that over time that is obviously obviously multiplies.
00:38:02.480 --> 00:38:21.519
Um, but I would love to shift gears a little bit and talk about design and where we're going with design, not only for interfaces, but how should companies start to look at design, not for the interface, for brand, in this evolution that we're going through with AI?
00:38:21.920 --> 00:38:22.480
Yeah.
00:38:22.800 --> 00:38:27.280
Um every company is extremely specific.
00:38:27.599 --> 00:38:35.679
They're born of a particular time and place with particular people building it in a particular way, right?
00:38:36.239 --> 00:38:59.440
And what I find the role of designers is is to kind of manifest that into being, which requires things like communication, requires uh prioritization, requires hierarchy, requires all this work to understand what should the thing be.
00:38:59.519 --> 00:38:59.679
Yeah.
00:39:00.000 --> 00:39:05.599
And how if you once you understand what it should be, then how do you actually manifest it very specifically that way?
00:39:05.760 --> 00:39:09.199
I'll give you a concrete example because that all sounds kind of like a little vague.
00:39:09.440 --> 00:39:16.880
But we have an AI for data center networking company that we just launched like two weeks ago, ARIA, right?
00:39:17.199 --> 00:39:24.719
And I can go uh into Claude or OpenAING, give me a bunch of names for an AI data center company.
00:39:24.800 --> 00:39:26.880
It'll give me all this stuff, right?
00:39:27.440 --> 00:39:34.639
Um, and many people in the company when we were doing it did that and came up with these big long lists of these things, right?
00:39:34.880 --> 00:39:38.400
But ultimately the name came to me when I was like out biking.
00:39:38.559 --> 00:39:42.800
I'm like, oh, ARIA, it's kind of like a symphony, it connects all these things together.
00:39:42.880 --> 00:39:49.119
And the founder, by the way, had a requirement that it has to start with the letter A, because his last two companies started with A and they were successful.
00:39:49.599 --> 00:39:50.559
Yeah, it's important.
00:39:51.039 --> 00:40:05.519
So there's like this role for the human there to really like pull all this stuff together, but then also have this kind of human moment where you're like, ah, that's you know, that's the thing that feels right.
00:40:05.840 --> 00:40:06.320
Yeah, yeah.
00:40:06.639 --> 00:40:21.599
And I think when a lot of people talk about, oh, taste is the differentiator going forward or judgment, yeah, machines can make something aesthetically pleasing and they can follow a trend of what's on you know on brand or something like that.
00:40:21.840
--> 00:40:26.960
But until they've lived a life, they don't have feeling.
00:40:27.679 --> 00:40:35.039
And that I'm distinguishing right like that from this concept of like, oh, this looks right a particular way.
00:40:35.599 --> 00:40:42.400
And I think the stuff that we resonate the most are the things that elicit particular feelings for us.
00:40:42.559 --> 00:40:45.760
Now, let me give you another nice example of this.
00:40:45.840 --> 00:40:49.119
And Jan Lacun used to give this example, but I I kind of like it.
00:40:49.199 --> 00:40:59.039
Like, you can teach a machine to drive a car by showing it thousands of hours of people driving, of a car being driven, right?
00:40:59.199 --> 00:41:00.480
And you train it that way.
00:41:00.639 --> 00:41:03.599
But the way we learn is we grow up, right?
00:41:03.679 --> 00:41:06.320
We grow up seeing cars when we're little kids.
00:41:06.559 --> 00:41:12.559
Like we're in a car with our parents, you know, our friends use cars to go places.
00:41:13.440 --> 00:41:18.960
It's just a very different way to gain the world knowledge of the car.
00:41:20.000 --> 00:41:23.840
Yeah, and the car to you, us as humans, has a very different thing, right?
00:41:24.000 --> 00:41:27.280
Oh, that's where I took my you know, girlfriend on the first date.
00:41:27.360 --> 00:41:30.960
We drove up in my car, or we went camping in that car.
00:41:31.119 --> 00:41:38.239
Oh, remember, like there's this, oh, I was able to finally drive and leave my parents' house on my own.
00:41:38.400 --> 00:41:46.239
Like there's this more, there's more to it than just the uh the mechanics.
00:41:46.320 --> 00:41:46.480
Yeah.
00:41:46.639 --> 00:41:49.920
And I think that's kind of the distinguishing factor for me.
00:41:50.079 --> 00:41:52.559
It's why I bring up that, you know, naming things.
00:41:52.639 --> 00:41:59.039
Like the machine can create thousands of names, but they don't necessarily have that, you know.
00:41:59.599 --> 00:42:00.719
Yeah, the emotional tie.
00:42:00.800 --> 00:42:16.559
I think that there's there can be lots of companies that come out and they just, you know, vibe code their way through a branding process, are but the companies that people will connect the most with will still be the ones that evoke an emotion, that have a story.
00:42:16.719 --> 00:42:34.800
People want authenticity and they want, they're gonna as and I think that that as AI enables folks to do that more, you know, create the brand and the in the interface all with vibe coding, I think people are gonna start to crave even more that connection and that journey and that authenticity.
00:42:35.039 --> 00:42:45.119
And so that's one of the things that we focus on a lot on Babco is is really setting the stage, you know, because there are gonna be companies who can be successful who are gonna vibe code their way there.
00:42:45.280 --> 00:43:07.519
But the companies, I want to get your thoughts on this, the companies that we believe will be iconic, that will be the new the new Nikes, the new Airbnbs, the prolific companies, will still be the ones who need to work above the LLMs to be thought leaders in the space and have that level of oversight and direction that is evoking the feelings and authenticity and the story.
00:43:07.920 --> 00:43:12.159
Yeah, I I think we as humans want things made by humans for humans.
00:43:12.800 --> 00:43:22.800
Not necessarily because those are the most efficient things or those are the most, you know, scaled or cost-effective things, but because we're human beings.
00:43:22.960 --> 00:43:23.360
Yeah.
00:43:23.599 --> 00:43:28.639
And like I again, it's it's almost like an intangible aspect of it.
00:43:29.199 --> 00:43:32.000
Uh the word feeling is the thing that I come back to the most.
00:43:32.079 --> 00:43:52.239
Like, I don't until again to come back to Jan Lacun's example, until you've like lived a life and gone through all those experiences, you can't really internalize the way that has an impact on what product you choose to buy or what you choose to do day to day or how you choose to do it, right?
00:43:52.320 --> 00:43:56.639
Like our whole this comes back to my kind of rant about context.
00:43:56.800 --> 00:43:59.760
Like every minute of your existence somehow.
00:44:00.159 --> 00:44:00.480
It's true.
00:44:00.800 --> 00:44:05.679
Pop something in there somewhere that later might be like really influential.
00:44:06.159 --> 00:44:09.440
You may like decide to open a door because who knows what.
00:44:09.599 --> 00:44:10.000
Yeah.
00:44:10.239 --> 00:44:12.559
Um the machines don't have all the context.
00:44:12.639 --> 00:44:14.559
And here is a perfect example.
00:44:14.639 --> 00:44:16.400
I tried to Waymo myself here.
00:44:16.480 --> 00:44:16.719
Yeah.
00:44:16.960 --> 00:44:17.840
I love Waymo.
00:44:18.079 --> 00:44:24.079
I sit in the back, I just relax, and I'm just doing my work, and Waymo's driving, and I'm like, Waymo, you're so awesome.
00:44:24.400 --> 00:44:30.800
And then today I called a Waymo, and I swear I was one second late from being able to be right there.
00:44:30.880 --> 00:44:32.559
And I was waving it down.
00:44:32.639 --> 00:44:32.800
Yeah.
00:44:33.039 --> 00:44:34.880
Because my daughter was coming out the door.
00:44:34.960 --> 00:44:37.599
She's like, Mommy, tell Luke he's gonna do awesome.
00:44:37.760 --> 00:44:38.559
By the way, she says that.
00:44:39.920 --> 00:44:44.800
And then um, I was I was literally a couple seconds late, and the Waymo just drove away.
00:44:45.119 --> 00:44:48.320
It did a loop around my cul-de-sac, it came back, and I stood right in front of it.
00:44:48.480 --> 00:44:49.360
I said, No, stop.
00:44:49.840 --> 00:44:51.119
And it was like, I don't know who you are.
00:44:51.199 --> 00:44:52.320
And then I tried to call a new one.
00:44:52.400 --> 00:44:53.760
It was like next Waymo 20 minutes.
00:44:53.840 --> 00:44:55.280
I was like, it's right here.
00:44:56.320 -->
00:44:57.519
So I drove.
00:44:57.760 --> 00:45:00.480
Um, but um on my own.
00:45:00.800 --> 00:45:04.079
But it was one of those, it was missing the context.
00:45:04.159 --> 00:45:07.199
It had no idea who I was, that I was standing right there.
00:45:07.280 --> 00:45:11.920
I had my phone, I had the app, everything was there, but it just missed that layer of human judgment.
00:45:12.000 --> 00:45:14.559
It was like, nope, don't care about you right now.
00:45:14.880 --> 00:45:15.440
See you later.
00:45:15.760 --> 00:45:18.800
Yeah, I hopefully some of those kinds of things I think get better.
00:45:18.960 --> 00:45:19.280
Yeah.
00:45:19.440 --> 00:45:19.840
Right?
00:45:20.079 --> 00:45:25.199
But again, the I keep coming back to Jan Lacun's thing.
00:45:25.280 --> 00:45:32.079
He's got like a different architecture for training these things, which tries to mirror more like the whole lifetime full of living.
00:45:32.239 --> 00:45:36.400
And maybe we'll get those kinds of systems soon, right?
00:45:36.719 --> 00:45:42.880
But um the ones we currently have, they can read freaking everything.
00:45:42.960 --> 00:45:44.480
They can watch everything.
00:45:44.639 --> 00:45:47.519
But like when they do, do they feel anything?
00:45:47.599 --> 00:45:47.920
I don't know.
00:45:48.000 --> 00:45:49.679
Some people might argue that they do.
00:45:49.840 --> 00:45:55.119
Like Anthropic has uh psychologists or something that evaluate the feelings of the LLM.
00:45:55.199 --> 00:45:58.320
So some people certainly think that's the case.
00:45:58.559 --> 00:46:03.760
It may just be like, you know, shadow, pseudo, you know, you kind of think you see it.
00:46:03.920 --> 00:46:06.400
We we we tend to anthromorphize everything.
00:46:06.559 --> 00:46:07.199
Yeah, right.
00:46:07.280 --> 00:46:10.159
So we're probably doing the same thing to these.
00:46:10.320 --> 00:46:18.639
But like I I have a hard time believing when the machine is going and just going through reams and reams of freaking data and it's reading all this stuff.
00:46:18.800 --> 00:46:23.599
It's I don't think it's having the same sort of reaction that we have to it.
00:46:23.679 --> 00:46:26.719
And I just fundamentally believe that reaction is really, really important.
00:46:27.599 --> 00:46:33.599
And as to come back to your question about designers, like our whole job is to give people that reaction, right?
00:46:34.079 --> 00:46:39.280
This is what I mean by like render things in the way that they should be intended.
00:46:39.360 --> 00:46:40.239
Um Yeah.
00:46:41.039 --> 00:46:46.400
So what what do you think in terms of we used to have design-led companies?
00:46:46.559 --> 00:46:51.760
Like I would say Airbnb is a design-led company, Apple is a design-led company.
00:46:51.920 --> 00:46:52.800
What does that mean?
00:46:52.880 --> 00:46:54.880
What do you think that means now with AI?
00:46:54.960 --> 00:46:56.480
And does that still exist?
00:46:56.800 --> 00:47:14.159
Yeah, if we go
back to kind of what I was talking about there, it's that thing that you are, that's in you, that you've built up over your entire life that governs what kind of decisions you make and what you choose to ship and how you choose to ship it, and whether or not you go in this direction or that direction, right?
00:47:14.320 --> 00:47:22.480
And some people are gonna be more design-oriented, some people are gonna be more engineering oriented, some people are gonna be more business oriented, right?
00:47:22.639 --> 00:47:29.119
But like ultimately, there's still humans making decisions, and some will argue that's not gonna happen in the future, right?
00:47:29.360 --> 00:47:37.039
But uh when that human makes those decisions, it's through that lifetime of living and they bring a perspective, right?
00:47:37.360 --> 00:47:43.199
Uh I mean Airbnb still does engineering, they still do business stuff, they still do legal stuff, right?
00:47:43.360 --> 00:47:54.320
So I think it's a little bit more around like the ethos and how they think about what they're gonna do than you know, like we only do designy things.
00:47:55.440 --> 00:48:02.159
What what are what are a couple things that you would tell founders in this space outside of what we talked about already?
00:48:02.320 --> 00:48:03.519
Use the technology.
00:48:03.679 --> 00:48:12.079
But if you're a founder in this space, you're creating any number of uh startup and any number of you know categories and industries.
00:48:12.239 --> 00:48:14.719
What are some things they can keep in mind right now?
00:48:15.119 --> 00:48:20.719
Yeah, I mean, we talked a little bit about the be mindful of what isn't possible but might be.
00:48:20.960 --> 00:48:22.800
That goes two ways, right?
00:48:22.960 --> 00:48:32.559
Because A, something could just happen that uh kind of nullifies what you've been working on, uh, or at the same time could come in and kind of enable what you've been working on.
00:48:32.800 --> 00:48:39.920
The types of things that we've been spending a bunch of time on are A, let's assume there's abundant intelligence everywhere.
00:48:40.079 --> 00:48:43.199
What do we need to support that abundant intelligence?
00:48:43.440 --> 00:48:49.039
What are the systems under those systems and what do they look like and how do they operate?
00:48:49.199 --> 00:48:59.039
Because that's a pretty obvious area that doesn't seem if you believe the AI stuff is here to stay, okay, what does it take to keep the AI stuff up and running?
00:48:59.360 --> 00:49:00.719
How do you make it better?
00:49:00.880 --> 00:49:02.239
Right, all these kinds of things.
00:49:02.320 --> 00:49:04.159
So that's kind of one area.
00:49:04.400 --> 00:49:14.320
Another area is to think through what are specific applications of this that run really deep across almost a full stack.
00:49:14.480 --> 00:49:19.760
So I'll I'll go back to the uh AI networking for data center company.
00:49:20.000 --> 00:49:26.480
Like they have optimized hardware that collects a particular kind of data.
00:49:26.719 --> 00:49:41.360
It gets uh understood and evaluated by AI all the way up to a layer where a human being can use AI to control and change it, but it's like a very full-stack solution in a particular area, right?
00:49:41.519 --> 00:49:44.639
Because that's not something necessarily that open AI would go do.
00:49:44.719 --> 00:49:46.000
Well, who knows what they'll go do?
00:49:46.079 --> 00:49:47.679
I can't, I guess I can't say those kinds of things.
00:49:47.840 --> 00:50:00.320
But right, but like taking kind of like a very full, almost integrated approach to something allows you to take advantage of the capabilities we have now in different layers.
00:50:00.559 --> 00:50:04.559
And so what a lot of companies will do is they'll just like focus on just the software piece.
00:50:04.800 --> 00:50:06.960
And that I think is a little bit more fragile.
00:50:07.440 --> 00:50:10.639
Because like, yeah, anybody can go make software right now, right?
00:50:10.880 --> 00:50:27.840
But if you're kind of going all the way down here and each piece is uh kind of expanding and complementing or dependent or whatever you want to say on all the others, then I think you have something that's a lot more durable and useful going forward.
00:50:28.079 --> 00:50:30.800
So those are kind of like the two things that those are good.
00:50:30.960 --> 00:50:31.280
Yeah.
00:50:31.760 --> 00:50:35.679
And what what do you think is like the white space to still solve?
00:50:35.840 --> 00:50:37.840
Because there's so much competition right now.
00:50:38.079 --> 00:50:41.760
Like there's so many things that can be built, so there's a lot of competition.
00:50:41.840 --> 00:50:45.760
There's a lot of companies doing different things, working in the same categories.
00:50:46.239 --> 00:50:49.280
What do you where do you think hasn't been solved yet?
00:50:49.440 --> 00:50:50.639
Where's the white space?
00:50:50.960 --> 00:50:51.840
I mean, everything.
00:50:52.000 --> 00:51:00.559
Like, so is the computer actually a an optimal representation of what we can do with technology?
00:51:00.880 --> 00:51:01.760
Hell no.
00:51:02.400 --> 00:51:06.880
It's like 90% amazing and 90% crappy at the same time.
00:51:07.039 --> 00:51:07.280
Right?
00:51:07.440 --> 00:51:20.400
Like still clicking on little icons and windows and opening apps and typing things in boxes and moving around and where's the folders and like it doesn't take a lot to think about, like, oh, that could be a lot better.
00:51:20.559 --> 00:51:20.719
Yeah.
00:51:20.880 --> 00:51:21.119
Right?
00:51:21.199 --> 00:51:36.480
And like this stuff, it permeates everything, every single kind of area of, you know, whether you run a restaurant, you're clicking on the thing, or whether you run a gym, or whether you run a software business, everybody's doing the same.
00:51:36.719 --> 00:51:41.199
Yeah, so there's a lot, endless amount of opportunities of
what to do there.
00:51:41.280 --> 00:51:43.199
Like let's just take the whole restaurant thing.
00:51:43.360 --> 00:51:48.880
Like people should be able to focus on the customer, right?
00:51:49.039 --> 00:51:52.400
And when you walk in, like, and I'm just idealizing, right?
00:51:52.480 --> 00:51:55.039
Like customer walks in, you should know what they like to eat.
00:51:55.119 --> 00:51:55.360
I don't know.
00:51:55.599 --> 00:52:11.599
But you can imagine all these different things that you could do that are uh, you know, a lot more natural, if you will, and integrated in the actual experience you're having, as opposed to constantly drawing people into little devices everywhere.
00:52:12.000 --> 00:52:31.039
Yeah, I think we have this, you know, we're in the Silicon Valley, we're in this little bubble where we're like, oh, you know, dev tools, design tools, and all these different tools that we create, but we forget, you know, that sometimes we can forget that there's all these applications to it that no one has started to use as much.
00:52:31.119 --> 00:52:39.360
Like for medicine and for restaurants and for real life industries that haven't been tapped into yet as much.
00:52:39.760 --> 00:52:40.960
Yeah, there's a lot of space there.
00:52:41.280 --> 00:52:42.960
What was I can't remember the name of the company.
00:52:43.039 --> 00:52:58.239
There's a company that put little uh collars on cows, and they have AI models that basically like figure out where all the cows are, where they go for optimal food, how they like just all the data of a farm, and it's yeah.
00:52:58.639 --> 00:53:03.119
That's a great we it's a very strange when you think about it, like weird application, right?
00:53:03.280 --> 00:53:05.119
But at the same time, it's white spaced.
00:53:05.199 --> 00:53:05.440
Yeah.
00:53:05.679 --> 00:53:17.360
Um my husband knows a a gentleman who owns Neo Tempo and it's a winery, and his whole thing is like creating and monitoring the vines and understanding what the each vine needs in real time.
00:53:17.440 --> 00:53:17.679
Yeah.
00:53:17.840 --> 00:53:21.280
And that's a perfect application that is is is white space.
00:53:21.360 --> 00:53:22.639
Like he's created something.
00:53:22.960 --> 00:53:25.840
But it probably still has like a little app interface and a website.
00:53:26.400 --> 00:53:26.639
That's it.
00:53:26.960 --> 00:53:29.519
With clicks, with clicking, with lots of clicking and all that.
00:53:29.679 --> 00:53:30.000
Yeah.
00:53:30.320 --> 00:53:34.960
If you could imagine in five years, what do you think?
00:53:35.119 --> 00:53:35.760
What is it?
00:53:35.920 --> 00:53:43.360
What's what what what are we what is the what is the team look like and what is AI look like?
00:53:43.519 --> 00:53:45.280
What is that vision for you?
00:53:45.840 --> 00:53:47.440
Five years is a long time.
00:53:48.079 --> 00:53:50.239
Okay, six months, one year?
00:53:51.119 --> 00:53:54.000
Yeah, may maybe in the next year, right?
00:53:54.159 --> 00:53:56.960
Because we're already seeing some of these things, right?
00:53:57.199 --> 00:54:17.760
I think the quality and capability of agentic systems matures where they can begin to feel a little bit more like teammates or at the very least, you know, highly capable resources that you can call on for specific things or set to do specific things.
00:54:18.079 --> 00:54:26.079
Um, and you can do those without you being there in the moment, managing, tracking, and those kinds of things, right?
00:54:26.159 --> 00:54:40.719
Like the autonomous nature, I think, begins to become more and more uh not obvious, but more and more apparent to people that like I can actually walk away, things are gonna happen.
00:54:41.119 --> 00:54:44.960
And I think that's gonna change the way teams work, right?
00:54:45.039 --> 00:54:49.360
It already is happening now, it's just not widely distributed yet.
00:54:50.079 --> 00:54:53.840
Um but I think that's a big shift that teams are gonna have to go through.
00:54:54.320 --> 00:54:57.920
It's not just about like what we're doing now, nine to five.
00:54:58.320 --> 00:54:58.639
Yeah.
00:54:59.679 --> 00:55:14.880
Well what one one last question here, and and I'm curious, like, what would you say to companies and founders who are competing against like what the capabilities are of like an anthropic and uh an open AI?
00:55:15.119 --> 00:55:24.000
Like, because there are you know, these companies they they they create something and then it's disruptive, and then you know, like you said, companies are like you either adds to your strategy or diminishes it.
00:55:24.079 --> 00:55:24.960
So, what would you say to them?
00:55:25.440 --> 00:55:32.320
It's the same thing that I would say back during the web or back during uh mobile or cloud or whatever, right?
00:55:32.480 --> 00:55:39.119
Like a strategy of is Google gonna do this or not is kind of an irrelevant point, right?
00:55:39.519 --> 00:55:48.800
Meaning, are you making something for people that actually helps them in a way that is appropriate for what they need?
00:55:48.960 --> 00:55:51.840
If you do, then you know, great things can happen.
00:55:51.920 --> 00:55:54.800
But if instead the way you're approaching your business is like, what's Anthropic doing?
00:55:54.880 --> 00:55:55.599
What's Google doing?
00:55:55.840 --> 00:55:56.880
Did that company do this?
00:55:56.960 --> 00:56:02.800
And you're watching like other startups and other companies, then you're focused on the wrong thing.
00:56:03.039 --> 00:56:08.639
What you need to be focused on is like, what am I trying to bring into the world and why?
00:56:09.039 --> 00:56:18.559
And when you focus on that, then you have a much clearer understanding of why you either are or aren't the right person or people to do that, right?
00:56:18.719 --> 00:56:26.559
Because it may actually be that anthropic is in a you know is the right people or person to bring X into the world.
00:56:27.119 -->
00:56:34.480
And what would you say to founders that decide that they look at that and they say, oh, actually it might be anthropic or open AI that's better at this?
00:56:34.559 --> 00:56:35.440
What what would you tell them?
00:56:35.840 --> 00:56:37.599
That's maybe a good decision then.
00:56:38.320 --> 00:56:39.679
Focus on something else.
00:56:40.000 --> 00:56:41.119
Maybe focus on something else.
00:56:41.280 --> 00:56:50.400
Yeah, I mean, it's it's not like ideas are that unique, is another important thing, right?
00:56:50.559 --> 00:56:54.079
Uh during the web, like how many e-commerce sites were there?
00:56:54.239 --> 00:56:54.880
Bazillions.
00:56:55.039 --> 00:56:57.039
During mobile, how many video sites were there?
00:56:57.199 --> 00:56:58.239
Bazillions, right?
00:56:58.320 --> 00:57:06.239
There's always like people that come to very similar conclusions because everybody can kind of look at and see the same things.
00:57:06.400 --> 00:57:10.239
But ultimately, it's not that short form video was a bad idea for mobile.
00:57:10.800 --> 00:57:18.079
It turned out to be a, you know, TikTok, and before that a little bit of Vine, and now, you know, all these other companies have kind of adopted it.
00:57:18.159 --> 00:57:26.079
But so it's the idea is a small part of it, it's everything else that is kind of the bigger part to it all.
00:57:27.119 --> 00:57:33.519
And we'll end on how do you stand out if you think that you are the right person, but you're competing against others.
00:57:34.159 --> 00:57:40.559
That's I've been saying for a while now we're in a golden age of building and a dark age of distribution.
00:57:40.719 --> 00:57:41.199
Yeah.
00:57:41.519 --> 00:57:50.480
Because attention is an even higher kind of premium, and we're actually seeing people do like outlandish things to get attention.
00:57:51.039 --> 00:57:55.840
And that doesn't just include tech, that like includes the whole the whole world, right?
00:57:55.920 --> 00:57:58.960
Is doing like crazy stuff to get attention.
00:57:59.199 --> 00:58:04.000
Um my personal approach to this has always been just consistency.
00:58:04.559 --> 00:58:05.760
Goes a really long way.
00:58:05.920 --> 00:58:16.159
Because you know, you can have a bunch of little shock moments here and there, but if you you know stay the course and focus, it tends to get better over time.
00:58:16.880 --> 00:58:19.920
But it is really, really hard right now.
00:58:20.079 --> 00:58:23.039
I think the easy part is like, let's here's the thing I made.
00:58:23.199 --> 00:58:23.360
Yeah.
00:58:23.599 --> 00:58:26.960
The hard part is, oh, people know about it.
00:58:27.280 --> 00:58:27.840
Yeah.
00:58:28.320 --> 00:58:35.920
People know about it, and they know about it here, they know about it there, they know about the context and the journey and all the information about it enough to make a decision to still use it.
00:58:36.480 --> 00:58:40.079
Yeah, that's a lot of things to get through to people.
00:58:40.559 --> 00:58:46.480
Well, Luke, this has been um, this has been packed, and it's been a pleasure to talk with you.
00:58:47.360 --> 00:58:55.199
And just to catch up and to speak on this level with you about interface design um and where you're seeing AI go.
00:58:55.360 --> 00:58:58.480
Like, I've I feel like it's been fruitful for me.
00:58:58.639 --> 00:58:59.119
So thank you.
00:58:59.360 --> 00:59:00.159
Yeah, my pleasure.