Subscribe
Share
Share
Embed
Conversations with designers, founders, and builders behind some of the best work
00:00:00:00 - 00:00:22:17
Unknown
I very nearly got fired before starting because I sent an email. Hey, like, actually, I think lovable should be a web browser. And I wrote down a thesis for like this is why this should this product should be a web browser before long existed. And Fabian replied to me and was like Majesty almost got fired before his first day for emailing his CEO a thesis on why lovable should be a web browser.
00:00:22:18 - 00:00:51:07
Unknown
He's a generalist who fell into design by accident. He wanted to build things, tried programing, you know, quickly realized that it drained the energy from his soul, but then discovered that people would pay him to solve problems a different way. Now he's the head of design at lovable, a company at the center of the AI coding explosion. I wanted to understand how design leadership works at a company that is democratizing the ability to build software when everyone can create.
00:00:51:07 - 00:01:20:00
Unknown
Who decides what's good when prototypes are free? What's the new bottleneck? And dad's answers surprised me. He told me their design system is now built for agents, not people. That the two most important stakeholders for vibe coding platforms aren't engineering and product. There's security and design. This is state of play. Let's get into it.
00:01:20:02 - 00:01:41:21
Unknown
The thing that you've been talking about for a while, and since I've known you is being a generalist, and you've kind of been I think you've been a generalist most of your career. Is that true? Yeah, 100%. I mean, I fell into design kind of by accident. I wanted to build stuff, and then I figured out that actually programing sucks the energy for my soul.
00:01:41:23 - 00:01:57:18
Unknown
But comparatively, I tend to be pretty good at the sort of problem solving part, and then kind of learned that people would pay me to do this design thing. And so, yeah, I've been a generalist like, since I was a kid. Like, basically what what about programing was like not was not not your thing.
00:01:57:18 - 00:02:20:20
Unknown
For me, it was debugging. I like to the creating part, but debugging just felt like speed bump, speed bump, speed bump. I think you, you're maybe a more advanced than me. I, I think I was just struggling with delayed gratification in general. I would be like, I want to, I want to get the thing. And it felt like I had to learn like years of frameworks and and then I'd also look at engineers who I admired and be like, they're amazing.
00:02:20:20 - 00:02:39:08
Unknown
So obviously I need to try to be amazing like them. And I just couldn't do it. I felt like I was like swimming against the tide. I felt like for a long time, the T shaped designer, gained a lot of traction. It was like, be a specialist in one area and then have kind of a little bit of breadth in, you know, some of the general other areas.
00:02:39:10 - 00:02:57:04
Unknown
Where would you say you were a T-shaped designer? Was there any area that you leaned into? Yes. I have a small anecdote here. And if, if Stephen, if you're out there and you'd listen to this by chance, you changed my life, actually, in a way that, you probably didn't realize, but, Dremel and Envision were on top, like.
00:02:57:04 - 00:03:16:12
Unknown
And they were just the absolute kings. There was someone called Steven Olmstead who was the community guy. He used to hang out on design and use and stuff like this, and I, I, him and I basically said, hey, I think I'm a pretty good designer. I've been doing this for about ten years, but I don't know what I should try to get really, really good at.
00:03:16:14 - 00:03:36:02
Unknown
And I got on a call with him and he was like, I realized early in my career that I'm a B-plus at everything, and there's nothing that I'm like an A star at. However, I'm really good at talking to people and doing all these other things. And then he ended up in like, community and advocacy and within design as an industry.
00:03:36:04 - 00:04:05:11
Unknown
And so I, I realized maybe halfway through my career that, I'm also not an A+ at anything, but I'm really, really good at 0 to 1. Very, very good at balancing craft and speed. And I also have a very, very high technical acuity where, I think I can sit at a table that design always can't when it comes to building super ambitious, technical products, essentially.
00:04:05:13 - 00:04:28:00
Unknown
And so I always index towards those. And I was completely not your person to build like a glossy mobile social consumer experience. Like, that's not me. But inversely, I tend I was pretty good at inventing things and designing things that just didn't exist before, and also just working with engineering to create, kind of, new technical approaches to problems.
00:04:28:06 - 00:04:45:18
Unknown
I started as kind of a cowboy designer, you know, like learning a lot of self-taught stuff, just trying to bring stuff to life. And I remember being in one of my first orgs and feeling like, because I was too much of a generalist, I actually felt like I sucked at a lot of things, and I was like, oh, I'm supposed to be like, in this box.
00:04:45:18 - 00:05:07:23
Unknown
I need to know how to do this a very specific way. And it wasn't until I, I allowed myself to kind of break out of working in a traditional org that I really started to feel like I had some sort of advantage. Who do you think today, with the way kind of work is starting to evolve? Do you think there are roles that used to feel like these?
00:05:08:01 - 00:05:46:04
Unknown
You know, they were meant for traditional org, but now are a little bit more powerful. Like, are there people today that would have probably felt like me in, in those previous organizations that maybe are more dangerous in an AI native environment? I think the people you have the gumption to kid an issue and then figure it out and then move on, and that can run things completely end to end, where you're going to go from having a user insight or an idea that you might sketch out on a napkin through to all the way to production, like, those are the people that, I see having the most impact and also the most fun as
00:05:46:04 - 00:06:14:07
Unknown
well, where, you know, you're literally it's limitless what you can do. So I think if you have those streaks, you're in a really, really great place. I think if you, if you subscribe to a methodology where everything you do has to be validated by someone else, be it data, research or other stakeholders, I think you're going to struggle, moving forward as as organizations change.
00:06:14:09 - 00:06:34:10
Unknown
And so and I also think teams will change as well around this as well. I like how you put that is gumption. You've got to see, I would imagine a certain shape of work more often than others amongst the people who are using lovable. Is there are there trends amongst the type of work you guys are seeing, or the way people are using lovable?
00:06:34:10 - 00:06:54:09
Unknown
Or is it just kind of is it really all over the place? Honestly, kind of all over the place. I can maybe split this into two. So like what we're working on as a team. We've got like a bunch of things that I would characterize as housekeeping. So easy example, we built like a workspace feature that's built to work with teams.
00:06:54:13 - 00:07:18:10
Unknown
When we built that, we were a 35 person company. So we paginate at like 50 people. Now we have enterprises with like 25,000 people. And it's like, how do you manage 25,000 people? And there's all this kind of readiness stuff where, we're frankly just behind. But the good thing is it's like easy SAS enterprise stuff and just no one's growing it, you know, the kind of scale and pace that we have before.
00:07:18:10 - 00:07:41:00
Unknown
But we can play catch up. But there's all this kind of readiness stuff that is I'd say, kind of like housekeeping. But then separately, we have a bunch of things we're working on where we're trying to figure out how do we cannibalize ourselves even, like, what's the next major product model breakthrough. That would change everything the same.
00:07:41:01 - 00:08:00:02
Unknown
The same way that, you know, felt like the world changed, two years ago, roughly. And that work is very exploratory. It's very much like, oh, let's delete and start again. Like, like what if there was zero users, zero traction? Like, how would we actually build a product starting again from scratch? And there there's a lot of 0 to 1.
00:08:00:02 - 00:08:23:13
Unknown
And like, a lot of, rapid R&D I'd say. Yeah. And you know, what's interesting is I mean I'm hearing from a lot of people because so many more roles, it's not just design, it's not just engineering but product even support, even sales are having the ability to kind of spin up these prototypes. And it's introducing, other challenges that I'm hearing teams are trying to navigate.
00:08:23:15 - 00:08:44:10
Unknown
And one of the things is now it is possible with tools like loveable, with all of the things that are out there, to have a sort of end to end ownership. Like you can really offer an idea and take that as far as at least the organizational let you to validate it. How are you guys seeing the definition of end to end ownership evolve?
00:08:44:12 - 00:09:12:11
Unknown
Yeah, it's interesting because, I truly believe that democratization is a good thing. Like you enable more things to more people to to do things and, you know, you enable more creativity, but in like a team context, you kind of can end up with spaghetti as well. Like there's this huge danger of because you can create something, even if you create it as a prototype, people anchor themselves to it and then suddenly, you know, your product can just turn into quite literal spaghetti.
00:09:12:13 - 00:09:35:06
Unknown
And so what, the way we're thinking about it and the way that we're seeing it is the designs role is shifting slightly, actually. And so like first of all, design has to be an enabler. We can't have this, you know, way of thinking of old where we would like data keep, waterfall decisions in some way being the holder of what good looks like.
00:09:35:08 - 00:10:04:20
Unknown
That just won't work. That secondly, the major goal of design now, for lovable or within our team is actually a goal around comprehension, where it's basically saying, okay, if if everyone can do everything, how do you have something that feels, comprehensive and singular as one experience? And so, funnily enough, I've completely understood on things like design systems, we didn't have a design system for the first 18 months, I'd say lovable, existing as a product.
00:10:04:22 - 00:10:29:08
Unknown
I've completely revisited my stance there. But the other thing that's interesting is where we've landed is that we tell everyone in the company, whether you're an engineer, like, whatever, if you're going to touch a product, you have the ability to be able to change things, because technically you can. But that ability comes with the responsibility to, to iterate extremely quickly if somebody taps you on the shoulder.
00:10:29:10 - 00:10:54:07
Unknown
And so like that means we have a culture of lots of people spiking things. But also killing darlings very quickly, deleting things, you know, twice as quickly as we spike things, in order to go for round two. And so, everyone kind of idiots and experiments and then kills things, and then we come up for a more considered, kind of, kind of second take.
00:10:54:09 - 00:11:11:00
Unknown
It's like one of the workflows that will undergo, I really want to talk about this team dynamic, because I get asked a lot from heads of design, you know, design leaders, even IQs who just want to be more proactive to introducing, new, interesting ways of working to the team. One of the first challenges I'm hearing from.
00:11:11:00 - 00:11:39:02
Unknown
So. So let's first look at people who are, like, already taken a couple of steps forward. They've adopted some new tooling. The they've they're given some room to kind of prototype some ideas and immediately when you start having a lot of people creating a lot of these ideas, you and I could have a 30 minute meeting. We'll jam up something real quick and it it informs the meeting, but that becomes this, like sort of institutional knowledge artifact that unless it's like captured somewhere, it kind of gets lost.
00:11:39:07 - 00:12:01:00
Unknown
How are you seeing teams or maybe how are you guys doing this where, you know, there's so much made now, but it's kind of floating in so many different places. How do you how do you start to capture the things that maybe non designers are creating that could be useful part of the process. Yeah. For for us we do we do a lot of like partial rollouts with control groups.
00:12:01:00 - 00:12:18:21
Unknown
And so if we think something has legs we'll roll it out a little bit and then just see what we can learn by doing that. And we try to do it in just the most reversible way possible of course, we're like way more measured on things that are difficult to change or, you know, might impact people's workflows materially and stuff like that.
00:12:18:23 - 00:12:41:03
Unknown
But if we just want to validate something, we'll just figure out what is the simplest slice of this, where we can at least give it to people and learn. And, you know, we'll give it to them quickly and easily in a controlled environment. And then we'll we'll kind of walk it back. And then we do that in production because I think it's difficult to get signal without the entire generative AI engine.
00:12:41:08 - 00:13:06:01
Unknown
And actually using it and touching it, like pictures of an AI engine. They don't cut it. But a lot of AI uses, they basically do the same with prototypes. And so what we're seeing is, a lot of very rich prototyping, going to, control groups, which isn't too big of a change, to, let's say pre I.
00:13:06:03 - 00:13:31:09
Unknown
But being honest, I completely underestimated how many people want to prototype that would never learn a prototyping environment. I have this huge blocker, this huge bias that because I knew how to use Figma or I need to how to write some frontend code or whatever, and I could test something. I didn't realize how constrained like marketers felt about like, hey, I want to test this messaging on this new product line.
00:13:31:09 - 00:13:57:06
Unknown
And, you know, my only tool is maybe doing some pay per click ads, and I can test out a couple of different page layouts to even see how I feel about my own copy. Or, a data person who's like, hey, like, I'm noticing that we get some drop off at this point of the funnel, and I'd love to even just think about, like, what would happen if we change this in order to have a richer conversation when I have a make a suggestion to a product team or like whatever.
00:13:57:07 - 00:14:19:12
Unknown
And it's essentially almost everyone but designers. And so and so, I had this huge personal blind spot that have been like spending decades actually wanting to build a prototype. And they just couldn't before because you had to let a canvas and then what tool noodling it is and all this stuff, just to make like the first version of an idea exist, which is crazy in hindsight.
00:14:19:15 - 00:14:40:05
Unknown
Okay, so I actually want to get nerdy here because I'm very curious about how lovable validates ideas. We we want people who have the gumption to just create. And I'm certainly I have experienced the overwhelm of being able to create anything and having to ask myself, oh, hold on, hold on, what should I create? You know, at what moment should I check in with myself?
00:14:40:05 - 00:15:05:19
Unknown
And then at what moment should I start checking in externally? So walk me first. First, let's explain to people who don't know what is the current structure of lovable. I mean, there's like roles that are pretty flat, but departments, how is that set up? Yeah. So with really broad strokes. So EPD engineering products and design who work on the core product, we have a bunch of products, engineers that are kind of split into different teams.
00:15:05:21 - 00:15:30:22
Unknown
One area is more AI forward, so builds AI engine and all the core interactions and makes it so that everyone else in the company can build an AI. Another area is, looks after the overall user experience using lovable. So dashboards, working with teams, collaboration stuff like that. And another area is more price focus. So like I said, like activation, conversion, new user experience and stuff like that.
00:15:31:00 - 00:15:54:11
Unknown
Within these areas we have teams. And so, every team has a mission. And then hopefully all these teams are complementary. Designers are embedded within them. And so designers have complete end to end ownership, like true ownership. Not every team has a product manager even. And so it's engineers and designers, figuring out what problems to solve, shipping things.
00:15:54:13 - 00:16:14:19
Unknown
Deciding where to double down on things, to iterate further, deciding where to remove things when they're not working. And so true ownership where, you know, my goal is I want every five person team to feel as empowered as when lovable was a five person company. Like, I want people to, like, truly act like an owner and within the context of that team.
00:16:14:21 - 00:16:42:13
Unknown
And then completely in parallel, we have a bunch of design rituals, where we critique each other's work and we try to make it feel like it comes from one singular team. And so like those two things kind of work together. But for us, the major, the major thing we index on is this reality and expectation of having true and so and ownership within the context of a team where you're working with the engineers.
00:16:42:15 - 00:16:59:13
Unknown
Okay. That's very helpful. So let's say then, you know, EPD, everyone's working, embedded designers are working on things. I'm on the marketing department, right? I come up and I say, hey, I've got a really interesting idea. We have a campaign we're going to kick off next quarter, and we want them to land on this landing page. It's really interactive in a certain way.
00:16:59:13 - 00:17:32:14
Unknown
What is the process, what occurs for us to say, yes, we should go ahead and ship that entirely if it's easily reversed or if we can easily iterate on it. More often than not, will I? Once it passes a quality bar, we'll just roll with it, and we try to make that as lightweight as possible. If we think there's going to be some kind of change management involved with getting this wrong, or if we think, it's risky, just just for whatever reason, we'll do some kind of partial rollout, and that's it.
00:17:32:16 - 00:17:55:19
Unknown
But actually, everyone in the company contributes to the code base, and so, right now we prototype with lovable, and so we take a version of lovable, and then we can riff on it, to come up with, like, testing links to prototype with. We can't publish all the way to production at the moment because, levels too big basically to build itself.
00:17:55:19 - 00:18:16:18
Unknown
But we're working on it. So I'll, I'll, I'll say the thing out loud that probably most teams that admit, like we use say, we use cloud code, we use all the good stuff. I mean, we use the right tool for the job. But we, we literally just try to we try to make sure people can go from insight to prototype, autonomously, essentially.
00:18:16:20 - 00:18:36:20
Unknown
So I build out this prototype in lovable. It's, it's got kind of the idea of what I want. And now is it that we just we turn a flag on for a percentage of the customers and they can see it already? Or is it that there are designated working groups of customers that you, thoughtfully put this in front of in like a meeting?
00:18:36:20 - 00:19:01:23
Unknown
How does that work? Yeah, it, it depends. So so what we've stopped doing, and this is like, you know, things that were acceptable in small team and small reach size but no longer acceptable. What we try to avoid is like yellow. The UI is going to change every day for people. And so, we'll, we'll try to do like a partial rollout with some kind of comms.
00:19:02:01 - 00:19:21:08
Unknown
And so, our default is we actually have a private slack channel with a bunch of beta testers where there's, I think it could like 1000 or so people that, at the moment, we also roll out a lot to our discord group as well, where we'll just ask people to opt in and then at least they've chosen to opted in that they're aware of the changes.
00:19:21:10 - 00:19:44:06
Unknown
And then completely in parallel, we also have and this sounds so enterprise we have our enterprise customary advisory board where we have we have the big guns. Here we show, some of our larger pieces to, every, every couple of months as well. Now you guys have you guys have gone through some serious growth in just the last 12, 16 months.
00:19:44:08 - 00:20:07:18
Unknown
At what point did you start introducing the advisory boards? When did that become necessary? It was when we realized, the extent that people were using bubble in the enterprise. Pretty much. And so we would literally have culls of people and we'd have, like an enterprise plan and be like, hey, coming soon, like we need to do, like we need to build.
00:20:07:20 - 00:20:38:04
Unknown
So we need to do like, like role based access controls, whatever we need, like a proper SLA. Please wait. But then we'd speak to people and realize that they were like, hey, I'm just going to put this on my personal card anyway. And so when we realized that was happening, and then, pretty soon after, people start doing that and, there's all kinds of security issues with where, like, we'd like to have a proper policy so that you know, that your data is safe, and we'd like to put the work in to do that.
00:20:38:06 - 00:20:56:10
Unknown
Started to become an issue. And so for us, it was in this is like six months after everything exploded. There was that, I don't know how vividly remember it. For me, it was like the winter, like the winter of vibe coding. And so it was as we entered the spring, it was like, hang on a second.
00:20:56:10 - 00:21:22:00
Unknown
I think we need to, like, catch our breath and just zoom out slightly and start to do things properly. Yeah, I've got to imagine. That's been crazy. Does it feel like it's slowed down for you, or does it feel like you've just gotten more deliberate? It, it feels like. Have you ever read about, how they try to temper traffic in LA by building more lanes on the highway and all that happens is that, like it says, age old problem, this fallacy.
00:21:22:00 - 00:21:40:21
Unknown
Like, you add a lane because you're like, oh, it's going to help the traffic, but then it just enables more traffic. That's how life feels, right now. Basically, it's like we add teams, we add people, we add capabilities to the product, and it just increases the amount, the amount that's possible. But it's good. It's like it's energizing.
00:21:40:23 - 00:21:59:04
Unknown
Which is I think the main thing for me. So the thing that I hear a lot from teams is, you know, some teams feel really fortunate. They're like the leadership and the the top down is making room. They're giving us budget for tools. They're kind of giving us a little bit of room for experimentation and learning those.
00:21:59:04 - 00:22:21:21
Unknown
Maybe there's designated blocks where someone can come in who's kind of a subject expert and teach us some things that maybe we're not aware of. How are you guys making room for throwaway work or experimentation? What is like how is that formalized it lovable? A couple of things. And actually everything I say, I'd say I'm also seeing this generally with like teams as well.
00:22:21:21 - 00:22:45:07
Unknown
And like, you know, especially the teams that each level that we talk to. So I think like internal hackathons and actually just embracing the fact that like, hey, we're going to spend company time on, things that may never see the light of day, but that that's not the point. The point is to, you know, spark a few embers and then and then we'll see if some of them catch, seeing a lot of that.
00:22:45:07 - 00:23:10:02
Unknown
And so we do that a lot internally. So just literally saying like sticks down, we're going to hack on something as an entire company or individual teams. Well, and very often the goal is like pick a different dimension to what you would do normally. And so let's like hack on something with either a completely different type of user in mind or a completely just type of capability or a completely different approach.
00:23:10:04 - 00:23:33:00
Unknown
And so like hackathons, I think have been like very, very enabling, for us. And also I see a lot of them, with teams, especially when there's like, you know, larger teams that are asking, still asking questions like should designers code? And so like, you know, things like hackathon and so like really great ways to at least, you know, let people experiment and help people play.
00:23:33:02 - 00:23:53:23
Unknown
So lots of that. The other thing that I think we do, I'd say like uniquely. Well, and I say this with some humility, as well, like, not that we're perfect, but I think a lot of companies will say things like, we want people to be empowered, we want people to be bottoms up, and they don't really mean it.
00:23:54:01 - 00:24:15:18
Unknown
And they'll say that, but they'll do a bunch of other things that, you know, end up strangling that in some way. What we do, I think pretty well, is we completely own what is top down. And so there's a bunch of things that just are top down. And so when it comes to, I don't know, huge big picture, how will the world look like in ten years time of successful?
00:24:15:20 - 00:24:34:05
Unknown
That is unashamedly top down. And actually we want it to be because like we're fortunate enough that like, you know, both Anton and Fabian, both of our founders have had incredible vision and it's and it's proved to be right, actually. And so what we'll do is we'll we'll try to own those things as top down and embrace them.
00:24:34:10 - 00:24:56:19
Unknown
But by doing that, I think it then frees everything else up to be bottoms up, if that makes sense. Whereas what I see a lot is especially when I've experienced as either when I see or a leader and, you know, some of my past roles in companies has been that we say we're bottoms up, but we never acknowledge what we're top down about, and then you just end up suffocating everyone by comparison.
00:24:56:21 - 00:25:15:09
Unknown
Give me an example of something top down. So like a mandate or just like, yeah, let's let's talk through that for a second. Like who, who we built for I'd say is very top down. And so we had a lot of, a lot of debate around, do we go for enterprise or not even just building for the web.
00:25:15:11 - 00:25:43:14
Unknown
So actually, I, here's a fun anecdote. I very nearly, probably got fired before starting because I sent an email, a week before I was going to go, full time at Liverpool. And my email was basically something in the spirit of, hey, like, actually, I think Liverpool should be a web browser and, I think it should be a web browser because the vision is to enable anyone to be able to create anything.
00:25:43:16 - 00:26:04:23
Unknown
And so the way that I'm thinking about it is it's like a browser, but instead of browsing the web, it creates the web instead. And I also think it's going to be better because it'll be desktop first. We can do things like run local databases. You'll have push notifications for workflows this, this, that. And I wrote down a thesis for like, this is why this should this product should be a web browser before long existed.
00:26:05:01 - 00:26:26:17
Unknown
And Fabian replied to me and was like, that's so dumb, we're not going to do that. And I'm so glad that he did. And so, you know, out of the back of that, there's a bunch of key decisions around like, okay, it's going to be web first and we're not going to worry about mobile. And we're going to be opinionated about tech stacks this way, and we're going to build around the models that way, etc..
00:26:26:17 - 00:26:46:07
Unknown
And I think it's been incredibly enabling that those kinds of things were set in stone, and that there was conviction behind those decisions, because if not, like maybe my argument would have won and we would have built a web browser instead and then who knows what have happened. You know, it's interesting to hear you talk about building a thesis around these ideas, these convictions that you've had.
00:26:46:07 - 00:27:02:19
Unknown
And I've got to imagine that for me, they've certainly evolved, and I don't remember when it was, but it was probably maybe 8 or 9 years ago with one of my startups where I sat down and actually wrote out a thesis like, where do I think this particular is live streaming? Where do I think this industry is today?
00:27:02:21 - 00:27:25:14
Unknown
What do I think it's going, and why do I think we should wedge on a certain bet? I wish more designers would do that. How often do you deliberately, whether it's, you know, writing it down or just like, really thinking about it and discussing it often the you kind of create thesis around your position. I do it pretty often, actually, and I even do it in my, in my like personal life.
00:27:25:17 - 00:27:51:13
Unknown
Like if I'm thinking about, you know, there's three different paths. And, you know, once we take, we take some direction with these paths. Is it going to be second, third, fourth, fifth order effects and there's going to be velocity. But behind that, that trajectory as well. I do it all the time. And so I'll like ask myself, like if everything goes right down this path, like what is that, what does that, lead us to.
00:27:51:15 - 00:28:09:15
Unknown
And I'll try to do that over, you know, different directions. So I do it all the time with products. Like, one thing we're thinking a lot about right now is mobile. So, you know, a lot of tension actually around. I'm sure you've seen that, like Apple, not allowing mobile apps that can essentially vibe code because they're worried about it.
00:28:09:17 - 00:28:30:15
Unknown
And so, you know, open question is like, okay, do we get creative and do we find workarounds because we believe in the web. And actually we think these kinds of products are in that good. And so do we like limit or do we just do something else. And so, you know, lots of internal chatter right now around something like that.
00:28:30:17 - 00:28:46:16
Unknown
But I also do it even with my like in my personal life with like where thinking about like schooling for my kid and so like where, like where are we going to live. And then that has an impact on our social life that has an impact on this, this, that or whatever. And so, you know, my partner hates it because she's she's way more yellow.
00:28:46:16 - 00:29:09:07
Unknown
And I'm like way more like I like to get to the philosophical creative truth. But I like for sure I like trying out different parts and then try and see if I can predict where the, where the land now, whether it's whether it's like a new thesis that you have, whether it's something someone on the team like, for example, I love a lot of orchestrating agents open claws, a place that I live and spend a lot of time at.
00:29:09:07 - 00:29:28:06
Unknown
And if I was on a larger team right now, I would be a great person. The knowledge share around that type of workflow. Whereas somebody else might have a better handle on something else. How does knowledge flow through the different? I sees it lovable. Yeah we do. We do a lot of like lunch and lens and so completely optional.
00:29:28:06 - 00:29:47:14
Unknown
It's in the company calendar. Like people just put them in for like Tuesday lunch. We're going to talk about this. So a lot of those things happen, organically. And I, I really love that they're purely cross-functional as well within the design team as well. We every Monday, we get all of our work out in the open as a team.
00:29:47:16 - 00:30:09:21
Unknown
And so everyone visualizes everything they're working on, and literally, like kind of flags as a team, we do around the room, like, who's underwater and so, like, you're just needs help with like, bandwidth but also who can, like learn and level up from someone else. And so, you know, as one example, I hate working in vector, and that's why there was no custom iconography the first eight months of lovable.
00:30:09:23 - 00:30:27:03
Unknown
And so, you know, that's an opportunity for someone to put their hand up and be like, hey, I'm horrible at this thing. And someone else can say, well, you know what? I'm pretty good. And so the interesting thing is, we started out with doing that for like, design critique, but we've ended up doing it with workflows.
00:30:27:05 - 00:30:53:21
Unknown
And so the way that we construct, like skill banks for agents now is like very much like, we're raising things like that more than we are, like actual, like visual artifacts for designs. And when this is a tangent, you're going to want to stick around. I can, I can sense it, but, like new design system is like 50% components, you know, of, like, here's building blocks for surface and like 50% skill banks, actually, as well.
00:30:53:21 - 00:31:13:12
Unknown
And so like half of our design system is now written for agents rather than, rather for people. Well, now so you start to introduce design systems and authorship. Who owns the documentation for design systems was always kind of a point of contention. Right. And it could very quickly get messy or drift away from the source of truth.
00:31:13:14 - 00:31:34:02
Unknown
How do you start to handle that with skills? A skill bank? Yeah. So we we experimented a lot and we had we had a couple of weeks of, like we called it like agent maxing. We were like, let's try and maximally use agents for like all of our design execution and see what works and see what doesn't work.
00:31:34:04 - 00:31:57:20
Unknown
What didn't work for us. And by the way, I'm going to enjoy listening to this in like, a year's time when everything's changed again, which would be great. What didn't work for us was background agents. So, anything where you wanted a person to really review it and really get down to the pixels and down to the metal, and that was going to happen in the background just didn't work because it felt like artificial productivity.
00:31:57:20 - 00:32:19:00
Unknown
You'd fire things off to like ten different background agents and then kind of wait and then have to context switch into like ten different results in ten different places and be like, okay, I could have just done this myself and really thought it through and done a great job. But where we landed on is we're seeing huge gains from having skill banks where we think of them like linters.
00:32:19:02 - 00:32:48:00
Unknown
So it's more like I'm creating something and I'm just going to like continually lint. And sometimes they just run automatically. They're like, oh, let's just make these like, hey, it looks like you're creating a component when an existing component already exists. That's not let's not do that. Or we call on them with specificity. So for example, we have one for keyboard focus, which has always been just something like really difficult to get right because you get it right in one view and then you get it wrong in the next one, and then it's suddenly inconsistent and you end up with clashes or whatever.
00:32:48:00 - 00:33:08:21
Unknown
Whereas now we can have like this keyboard focus skill that, you know, looks at our entire application application wide and just gets that correct. And so this kind of skills, this linter thing is working very well for us. And the other thing that's working really well is also using these as a part of, like pull request workflows.
00:33:08:23 - 00:33:32:12
Unknown
So as a part of like CI and CD, we have like automated testing, automated like linting, flagging, visual defects, stuff like this. And like these two things together, are working extremely well. And then taking that back to your actual question, the documentation feels like making those skills truly great and making them produce outcomes like design outcomes that we genuinely like.
00:33:32:14 - 00:33:53:22
Unknown
So, you know, I think we have I was trying to think of this before our conversation today. I think I have five internal tools that we use pretty regularly that we've built or that I've built that, that say, okay, we can use these things for, for our studio. How I've got to imagine lovable has quite a few handcrafted internal tools that keep things running.
00:33:54:00 - 00:34:16:04
Unknown
Is that true? Do you guys you guys use a lot of your own internal tooling? Like even just for for like the skill linting, the skill bake, there's maybe there's is there a tool that helps someone onboard to the skill bank or maybe like that houses them, or a dashboard for viewing them? Yeah. It's it's crazy. We have like, I'd say we're like, eating our sass stack, like, generally.
00:34:16:04 - 00:34:38:03
Unknown
And so like, like two anecdotes in, like, two different directions. One is we have an agent that we built unlovable, which is basically replacing all of our products, analytics and, like data warehouse, where, you can basically go to an app and talk to this agent and it can visualize things, or you can tag it in slack as a slack.
00:34:38:03 - 00:34:57:12
Unknown
But, I only learned this like a month ago. You can literally build a lovable app and say, like this app, a slack, but then it just will. And it's like magic, which is crazy. And so like, use cases like that, we're actually offloading from, some pretty like, proven data platforms that we were trusting to get insights.
00:34:57:12 - 00:35:18:09
Unknown
And we just don't need to anymore because we can just build something custom and then just get the answers. We want in the way that we want. And also just things like, okay, so you've managed people all HR software sucks. You know, when you want to plan different teams and every single piece of software forces one person to have one manager.
00:35:18:11 - 00:35:45:19
Unknown
And from a, you know, line manager set, if that's fine. But often people contribute to multiple teams and you're often thinking about like, okay, what is my roadmap? Or maybe this person will be like 50% time here, 20% time there. I need a dotted line from this person because they'll pay really well with this person or whatever. And you just can't do that with air software, like how HR software is built around like, you know, this person lives in this state and pays taxes there and reports into this, but so then that's it.
00:35:45:21 - 00:36:06:06
Unknown
And so we built this like custom HR software that lives on top of our like, you know, compliant, compliant, like proper thing that makes sure everyone gets paid. But it visualizes things like org charts on top. And so when it comes down to like, hey, we can I look at the roadmap and we're going to look at the things we want to do and the people we need, and we're going to look at hiring.
00:36:06:07 - 00:36:23:00
Unknown
We can almost have this like infinite canvas experience for HR and like that's where we can actually ideate across different planes and different dimensions. And there's like things like that where like I look back in hindsight and I'm like, okay, a year ago I would have solved that problem with like stickies on a whiteboard and or something else.
00:36:23:00 - 00:36:52:12
Unknown
I don't even know how it is. And I, I'd have done it manually, and now I can just create software to do exactly what I want. Yeah, what's interesting is I find myself interfacing with all of my internal tools. The starting point is slack and acts. I have different channels. It is, you know, whether even when I'm interfacing with like gusto, for example, and I have to do something like if I there are some software that I still have to log into because they simply don't have an API that plays, you know, well with what I'm trying to do.
00:36:52:14 - 00:37:15:12
Unknown
And, you know, that feels like it's very unique to us, that we are building the connected layers, but I don't really want to use the tool. And eventually, if I can completely off board from the tool to build that part out, I will. But my question is, how are you seeing that sort of thing happen at scale across your customers, or is that still seem to be like a very small fraction?
00:37:15:15 - 00:37:44:01
Unknown
I think in smaller teams. I think smaller teams are like moving quite quickly here. I think large companies are more risk averse. And especially you come down to this kind of build or buy decision and build also includes like compliance risk as well. And you're like, well, actually I'd rather pay the HR platform 100 bucks per seat per month because I know that I'm like, you know, I've ticked every box, from a compliance perspective or like whatever.
00:37:44:03 - 00:38:12:15
Unknown
And so like we're seeing that, the really interesting dynamic, just like more generally that we're seeing with all of this is, a platform like lovable lands into the enterprise. Everyone actually fights with each other over who owns Vibe coding because everyone wants to own it. And, you know, you have engineering saying, well, technically this is building, but then you have like marketers saying, well, I've wanted to do this for years, so I want to own it.
00:38:12:15 - 00:38:40:10
Unknown
I like, like whatever. Like literally people fight with each other, but the two stakeholders who appear very quickly and whose opinion really counts, which is an expert, is the first one is security, which is kind of predictable, like people are okay, what's happening with the data? This has to be secure. But the second one is design, which was a real surprise to me, actually, where what happens very quickly is people start creating lots of stuff and it's very obvious where it's badly designed.
00:38:40:12 - 00:39:05:13
Unknown
And so immediately it's like, hang on a second, like, this doesn't feel on brand. This isn't using our design system, whatever. And so actually the two most like key stakeholders for like five coding platforms at enterprise at the moment, if you ask me today are like security and design, which is interesting. It's not engineering. It's not like anyone else when it, when it when people start to argue over who should own, you know, the artifacts of vibe coding.
00:39:05:15 - 00:39:37:07
Unknown
How are you guys handling that. Yeah. So we we have a team kind of built around this problem. And so they're basically spending a lot of time like physically in the room with like, you know, really large customers. And basically making sure that, making sure that they have what they need. And at the moment, as I was saying, earlier, we have a lot of, like, readiness where, like, like we've just grown at a scale that was, unexpected, where there's a bunch of just housekeeping to do.
00:39:37:09 - 00:39:56:01
Unknown
But also this these kind of security and design themes are like the two, the the two sort of major focuses for us as well. At the moment. How is internal tool adoption handled? If someone, you know, if someone says, hey, I created a skill bank and I think like this would be really useful for those of you vibe coding, you can get our design system skills over here.
00:39:56:01 - 00:40:18:09
Unknown
That'll make this look a little bit more, like what? What our standards are. How does something like that get approved and then get kind of spread throughout the org? I can answer this both for lovable like for us as a team. And then I'll say it for like what we're seeing with teams as well. So so at lovable we I actually don't have a great answer for you.
00:40:18:09 - 00:40:34:14
Unknown
It happens by osmosis. Pretty much like people create great skills and they get great results. And other people are like, how did you how did you do that? And then they just they just sort of, start using them kind of very organically. And so I think we just have a curious culture which is, which is a good thing.
00:40:34:14 - 00:40:58:01
Unknown
I don't think we're doing anything too, too magical. But no, but no hard gates. No hard gates, I create something. Hey, look what I did. I can share it in slack. Someone else can use it. Free flow. Exactly. Completely open. Actually, almost everything is open internally. Like, I almost don't have, like, DMs. Even, like, if someone DMs me, I often DM them back and I'm like, this, this could have been in public.
00:40:58:03 - 00:41:23:14
Unknown
And then more people can benefit from it. So yeah, everything's in the open. Nascent the other trend that we're seeing with teams is that, one person create something, and there's almost this, like, product led growth mission that happens kind of internally where effectively, like apps and tools kind of go viral, where like one person creates it like maybe one team on boards.
00:41:23:16 - 00:41:39:00
Unknown
And like word of mouth seems to be, one of the biggest drivers here where, once you create something and then suddenly every other team gets FOMO and it's like, I want to use this tool. But B, how did you build it? And then they want to go and create their own stuff as well.
00:41:39:02 - 00:41:58:12
Unknown
So the kind of culture that you've you've explained on how lovable works, I think is a really interesting culture. And someone like me ten years ago would have loved to be a part of that because it just it seemed so freeing. It didn't seem like I would run into boundaries, in a way that was discouraging to me.
00:41:58:14 - 00:42:16:05
Unknown
But today there's a lot of designers either junior who have no idea what to expect from the market right now, or seniors who came from a lot of what we've gotten used to over the last 15 years. What would you say to both of those groups of people who are saying, I'm hearing I need to be more AI fluent.
00:42:16:06 - 00:42:43:22
Unknown
I need to be more AI native? What does that actually mean? And when you when you were looking at candidates, how were you evaluating AI fluency? Yes. So I mean, I fluency is like is paramount. I'd say, and for us just given, you know, where where we build, where we play, like actually we look for like AI maximalism, like, not just not just I've tried AI a little bit and I use it in this small part of my workflow.
00:42:43:22 - 00:43:04:09
Unknown
It's like, you know, treat AI as a material, because you're gonna have to shape it. I mean, so you have to understand how it works. And the same way that an architect must learn the laws of physics and you know how buildings are constructed, like we basically think the same with designing for AI. And so for us, we have a very high expectation of like AI fluency.
00:43:04:11 - 00:43:28:19
Unknown
And that includes how the models work as well. So very, very high expectation of fluency. And for maybe for the people who are like struggling or like not using it every day, I would do two different things like which are maybe the the same answer actually, but I would probably be a founder. Actually, I don't think there's a better time to be to be a founder.
00:43:28:19 - 00:43:49:04
Unknown
And I think actually as a designer, you're incredibly well placed. Because you can figure out what people want, you can build it with the help of AI, and you can shape it exactly in your vision, and you have all of the tools to be able to do that. And you can create something that's differentiated because you're a designer.
00:43:49:06 - 00:44:06:17
Unknown
So I would think about being a founder, if you don't want to go full sass, if you don't want to go full company, whatever, I would still be a founder, but a founder of some side projects. And so I would just scratch an itch and just see what falls out of it. And I would just treat it like play.
00:44:06:19 - 00:44:23:20
Unknown
It's like it feels like building with AI and designing an AI really reminds me of playing like an RPG or something where you're like, okay, I'm, I'm going to start at level one and I'm going to do this small thing and then it's like a muscle where you get the reps in and you can build, build the strength over time.
00:44:23:22 - 00:44:48:05
Unknown
And so I just start and, and just treat it as play. I love that treat it as play. It's something that I've, I've, I've said a lot. And it's been really helpful. But also on top of that, it's one of the things I've heard from a lot of teams is one of the first indicators. Oh, I know right away when I see somebody sharing a thing they made through play, that, you know, that qualifies them for maybe the next round.
00:44:48:05 - 00:45:05:21
Unknown
If we're looking for someone who's AI fluent, what other signals do you look for that tell you right away before you get into like the meat of the interview cycle, that somebody might be a good fit? I actually like to learn how people got into design, and I also like to learn how people learn as well. Like I'm completely self-taught.
00:45:05:23 - 00:45:28:19
Unknown
And so I learned everything from like literally right click, inspect and copy paste into notepad XY and then say see what happens when I save to desktop. Or I learned like visual design from like literally getting like a magazine cover and being like, I'm going to try to recreate this like I want to. I want to see if I can get that from the real world onto my screen somehow.
00:45:28:21 - 00:45:49:15
Unknown
And so I look for a lot of, like, self-learning, just generally. And it doesn't even have to be within design. Like, if somebody, I don't know if they link to their account or Instagram account from their resume and it turns out they're really into like woodworking or like whatever, I'm like, great, let's let's get nerdy. Like, this is a really, really good thing for me.
00:45:49:17 - 00:46:12:06
Unknown
And so like, like self-learning and like intrinsic drive, I think the two main things, one of the things that I and I love, that I want more people, I want to kind of underline or highlight the the self-learning, the intrinsic drive. I call it the love for the game because I really don't think you, you explore this space unless you really have something about it.
00:46:12:08 - 00:46:31:11
Unknown
It just pulls you towards it. Right? Doesn't mean that you have to be, you know, working Monday through Sunday. But I think for a lot of people who aren't experiencing a lot of overwhelm, it's because there's a lot of excitement to kind of it. What's under this rock? Hey, what happens if I push this tool just a little bit further with the teams that you guys work with?
00:46:31:13 - 00:46:54:05
Unknown
You work with a lot of teams, a lot of product teams. But I, I've also heard there's some really interesting examples of teams who are using lovable in some kind of non-conventional ways. Are there any that come to mind? I'm very curious to hear about this dude. So I, I worked in gaming for seven years, designing like UI for first person shooter games.
00:46:54:09 - 00:47:22:13
Unknown
And there was one studio that I always had as my, like, visual references. When I was doing this called territory, who were based in London, and they do every sci fi UI you can think of where anytime you've watched a film and you've been like, Holy crap, that just looks insane. Whether it's, you know, any of the Marvel films of like, the Iron Man had or like Alien Prometheus like Dune, you pick it, you name it, they've done it essentially.
00:47:22:15 - 00:47:52:00
Unknown
And they are using Liverpool to, to prototype and build UI essentially. And so they're using it for I have to watch my words because they're, they're a little bit careful with the clients. But but they're basically building custom design tools for these workflows. And so they're able to build these, you know, these like futuristic UI's and futuristic things, that we then see in like, film and television, which is crazy.
00:47:52:02 - 00:48:16:18
Unknown
That is, that is wild AI. That is the story I would love to tell the, I, I love hearing about those those kinds of things. I've been seeing all kinds of people building in so many different ways, especially in gaming. By the way, if we have a video coming out talking about like this Golden age of indie gaming, that we may see in the very near future, the, you know, the last thing that I want to ask you about, we were talking about these kind of theses that you've had.
00:48:16:19 - 00:48:38:14
Unknown
Is there is there you've you've watched this. I mean, you've been you've had a front row seat to see where Vive Coatings gone. We have the State of Prototyping survey that concludes this week to see how our teams actually doing this. Because there's so much social media hype, what do you actually think is the truth of what's going to happen around how software is made and how these companies kind of prioritize these roles over the next?
00:48:38:18 - 00:49:01:19
Unknown
I'm not going to ask you to pick a timeline, but what is your current thesis around that? I think in the future everything will be interoperable. And so I really hope we leave this world of like, okay, in this design environment, you're constrained to only do things in this way, and this coding environment is only accessible to people whose brains work that way.
00:49:01:21 - 00:49:21:05
Unknown
And so my big kind of bet directionally is that everything will work together in the future. And I think that's a really, really good thing. And I think the same way that, like, developers have been able to choose the development environment for decades from everything from a command line through to an ID through to 20 different flavors of idea.
00:49:21:07 - 00:49:40:23
Unknown
Actually, hopefully everyone on the planet can create software and they can also choose how they want to do it. And my hope is that it happens in a completely open way. That told me what he's looking for in designers said it has nothing to do with tool proficiency or even AI fluency that instead it's all about the ability for self-learning and intrinsic drive.
00:49:40:23 - 00:50:01:15
Unknown
And I think that maps to something bigger, because the best designers I'm meeting are the ones who can't stop tinkering with what's in front of them. Their curiosity persists even as things change, he said. When vibe coding lands inside of a big company, everyone starts fighting over who owns it and reiterated that security and design are the two stakeholders that keep showing up.
00:50:01:17 - 00:50:16:11
Unknown
Design went from gatekeeper to cleanup crew to maybe the quality conscience of an organization where everyone can build, and now seems to think that's an opportunity. And I think he's right. That's the episode. I'll see you next time.