We launched Lovable in November 2024 with a visual identity I prototyped in 30 minutes.
I joined eight months earlier as the fourth employee and first designer. I barely remember the weeks leading up to that launch because everything was happening so fast. Then the product went viral, and things got even hazier. We were racing to keep up with the $10M ARR we’d picked up in just 60 days.
What still sticks out to this day aren’t the business metrics (I had to go back and check the records for that). The moments I remember most viscerally were when I could feel the emotion from people discovering Lovable.
The Lovable brand we launched that November wasn’t supposed to scale. When we went viral, I immediately knew we needed a visual identity that could scale. So I talked to every paying user I could, trying to understand what Lovable made them feel. Rewatching those calls afterwards, I was struck by their body language — they sat up straighter, their eyes widened. When I met them in person, they’d pull my laptop out of my hands just to show me what they wanted to build.
I recognised the feeling. Using Lovable for the first time felt genuinely infectious. It reminded me of how magically expansive the internet felt when I was a kid, like anything was possible.
That was the feeling I wanted Lovable’s brand to communicate. I translated it into a design brief, and just a few months later, we rolled out our new identity — grounded in a vibrant pink-gradient heart turned on its side, quite literally a symbol of the love that goes into everything we make.
In just two and a half years, I’ve designed (and redesigned) the product, shaped the brand by following my own instincts and scaled an insanely talent-dense design team, all possible because Lovable’s co-founders Anton Osika and Fabian Hedin let me run with it.
Lovable has shown me that when people with innate curiosity and genuine care for what they do are given complete trust, they produce exceptional work, remarkably fast.
Finding the product I wish I had as a kid
I fell in love with design long before I knew what it was. Growing up in Leeds, England, in the ‘90s, I was obsessed with three things: music, gaming and the internet. I spent my free time building websites for local musicians, online communities and the semi-professional Counter-Strike leagues I played in. I’d stow away in my room to scan vinyl artwork on our A4 flatbed scanner — the sleeves were too big to fit in one pass, so I'd scan each record twice and stitch the halves together in a cracked copy of Photoshop, lining them up as precisely as I could.
Since I’d spent my childhood hacking around on the internet, I wanted to learn how to code so I could build the ideas I had in my head. But I quickly found out I hated programming — it didn’t come naturally to me, and it felt like it sucked the energy out of my soul.
In my teens I realised I could get paid for designing things online. I went from building websites for friends to building them for local businesses, which eventually landed me a gig leading ecommerce for the UK's largest music retailer at 19, while still at university. I'd enrolled in the first design course in England to include the internet in its syllabus, but dropped out in my second year. The tipping point was Gmail's release, and with it, Ajax — the technology that let websites refresh without reloading the page. When I was banned from using it in my coursework, I decided traditional education would never keep pace with technology. I dropped out and never looked back.
From there I worked my way up in tech, eventually landing at a scale-up in London called Element, where I became VP of Design. I never fully identified with the glossy, Design-with-a-capital-D leaders you'd expect to find at consumer social startups — I always preferred collaborating with engineers building technically ambitious things, even though I never wanted to be one myself.
But even as a more technically minded designer, I have strong opinions about brand. I’ve always thought that what we choose out of a lineup is the thing that’s made with the most care.
I credit that to my mum. My dad passed away unexpectedly when I was five, and she raised me and my brother on her own while working full-time. We'd spend weekends with my two aunts, who kept us busy by putting us to work in the kitchen. Cooking is an expression of love in our family, which has South Asian roots, and every Saturday I'd help with the meticulous prep work that went into a big meal — things like picking every single coriander leaf off its stem for hours. I saw all the layers that went into a thoughtful, flavourful dinner: 30 minutes on the onions, then the dry spices, then the fresh spice. As a kid, I understood that you could taste when a meal was made with love. I think that stayed with me.
So that was me before Lovable: a computer nerd who found a career in design as an extension of creating stuff on the internet as a kid.
At the start of 2024, I’d been spending nights and weekends playing around with GPT-3.5 at Element, obsessed with the idea that computers could now “think.” My team served customers in government, enterprise and public sector, and we had to do a lot of repetitive accessibility work to comply with stringent standards for web, iOS and Android. I figured out how to get GPT-3.5 to handle 70% of that work and became fixated on finding AI’s edges.
While exploring different tools to help me do that, I found an open-source project called GPT Engineer that had racked up 50,000 stars on GitHub. I messed around with it and was impressed, but I couldn’t tell who’d created it. I found a website for GPT Engineer, but it was so poorly designed I had to check the DNS records to figure out if it was made by scammers. I found Anton’s Hacker News profile, which said he was a Y Combinator founder, so I figured he was at least pretty legit.
I was hooked by this seed of what would become Lovable. Every hour I’d think of something to try that I didn’t think was possible with AI. It revived my long-held belief that the best products should make you feel something. I knew I wanted to be part of it.
Trial by ownership
I cold emailed Anton offering to help with design work, and soon enough we’d worked out a remote contract role (he and Fabian are in Stockholm, and to this day I live in London). I started with a trial of two days a week for two weeks.
On my onboarding call with Fabian, he didn’t mince his words: “Here’s a list of outcomes I’ll be grading you against. It’s all ambiguous. Create the structure yourself.”
He gave me a one-pager with three problems to solve:
- Make our crappy UI/UX and web app consumer-grade
- Figure out which customers this product should be for and scope a first version
- Build a community strategy — we have a GitHub and Discord community but we’re not sure if they’re our long-term users
I had just days to make measurable progress on each problem. My first week, I did a teardown of the product, website and onboarding to create a more intuitive experience, and I interviewed the internal team to shape a hypothesis on who this product was actually for. My second week, I redesigned everything: the brand, the website, our core UX, and by the end of my fourth day, I was shipping.
These were arguably jobs for three different people, and tasks that would take talented ICs weeks to chip away at. But I was energised by the scope of the ambition, and the trust they’d put in me to just figure it all out in the span of days.
The test of joining Lovable is being handed a preposterously large problem and seeing how you respond.
My first call with Anton drove this home. He’d been in San Francisco at the time trying to hunt down information from OpenAI about the next GPT launch. I wanted to sense-check whether he agreed with my approach to solve those three problem areas. I had started to explain one of my ideas, and he interrupted me to say, “If there’s one thing you should know about me, it’s that I want you to take action on the things you’re confident about automatically, instead of telling me first. I get annoyed when people don’t just do things on their own.”
I never asked for permission again.
I used to think the stakes of the decisions I had to make were high. I was setting the direction for foundational UX choices — and in a pre-AI world, getting it wrong meant time and money you couldn't get back. But I realised the game had changed: In a world where you could build and rebuild in hours, sunk cost stopped being a real cost. If we were wrong, we'd just try again once we had new information.
We called this clock speed: your internal pace of thinking, decision-making and executing. Some people's clocks tick faster than others. Anton and Fabian are fast, and they want everyone around them to be fast, too — which is what made me realise my organic clock speed is fast as well. At my previous job I'd deliberately slowed myself down: As you rise in seniority, you’re taught to think and communicate more slowly, taking care over how your words land.
For example, the original web app used a lot of red as a highlight colour. I wanted to reserve red for destructive actions only. I remember raising the change with Fabian and making the call in literal seconds.
Lovable pushed me to reset a clock speed I'd been suppressing for years.
After a couple of weeks of contracting part time, I flew to Stockholm to meet with Anton and Fabian in person. At the time it was just a couple of us in a non-descript coworking space in Stockholm, often staying late into the evening. I’d go straight to the office and back, so I didn’t even get to see much of the city. That was partially because I’d booked an absolutely awful room in a hostel to stay in because I wanted to be frugal. After I arrived, Fabian made a throwaway comment about how great it was that I didn’t waste money on expensive hotels (“Spend company money like it’s your own” later made it into the employee handbook).
I found Anton and Fabian a bit crazy, but in an endearing way. That impression of them still holds to this day.
They offered me a full-time position. I knew taking it was a gamble: My partner and I were expecting our first kid, and she had just lost her job the same month that I started talking to Anton and Fabian. But I couldn’t resist the pull I felt from the product, or the possibility of building it at a speed I hadn’t been allowed before.
Trying to convince the world (and ourselves) that Lovable could work
The summer before launch was an exercise in intuition. We spent our days wrangling with existential questions and building and rebuilding the whole app, because it only worked part of the time.
The company was just two technical founders, me, a PM, and a handful of engineers. We’d fully deleted and rebuilt the product twice in four months — each time it was incrementally more reliable, but the LLMs would still return errors roughly one out of every four prompts.
At the time, the default response to LLMs was to point out when they made mistakes, and it was hard to argue with the scepticism. As part of my work to figure out who should use Lovable, I got on a bunch of user research calls during this time — most thought it was vaporware, or a really dumb idea that only worked a quarter of the time.
Anton hadn’t started GPT Engineer with a specific user in mind, and none of us yet had the conviction to say who it should be for. It was only after it continued to get more traction that we noticed that non-technical people were finding a lot of value in it. That made us realise that AI code generation for the 99% of the world who aren’t professional engineers would be far more impactful.
This presented a core design problem: What’s the best coding tool for someone who doesn’t know how to code? Should they see the actual code or not? Cue an extended early debate about whether or not we should show code in the interface. The engineers trusted the LLMs more when they showed their work. I felt that non-technical people, the users we were considering building for, wouldn’t gain much value from reading code if they don’t understand it.
But we held our opinions loosely. We ultimately decided to move forward without code because I made the case that it was a two-way door — it would be easier to start without and add it later than the other way around.
So we rallied around the mission to build for the 99% and moved forward with a product that, for the most part, didn’t show code. But here’s the funny thing about intuition: I don’t think we were 100% sold on the “build for the 99%” mission before we launched. In all honesty, if the launch wasn’t as successful as it was, there’s a good chance we would’ve gone back to the drawing board.
The launch
Claude 3.5 Sonnet’s release gave us a technical breakthrough that allowed us to fully rebuild Lovable for the third time, and its reliability improved dramatically. After we’d finished V3, I got on the phone with users again, and this time they had a different tune. People lit up when they talked to me. They were suddenly able to do things they just couldn’t do before.
Two calls stand out. One was with a healthcare worker who told me about all the ways they wanted to use Lovable to make their team more efficient. They were so animated when they talked about their ideas. I remember being struck by the fact that we weren’t just enabling programming for the vague idea of vibe coding — we’d reached the 99%, and in this case it was real specialists who could push forward their field with this tool. Today, that’s our core user: the domain expert who is closest to the problem for which they can now build the solution.
Another person I remember was an Australian guy who has ADHD. He told me he’d tried every productivity tool out there, and nothing had ever worked for his brain. Every note-taking tool, every calendar app. So he started to build his own personal suite of software specifically designed for him.
Meanwhile, that autumn was a season of intense change in my life. My son was born, and I spent several weeks with him in the NICU monitoring heart issues (aptly, he was born with a condition that was generating too much electricity in his heart). I found some comforting escapism in the intensity of the work I returned to in the lead-up to launch. I figured out the exact position to rest my son on my forearms while I worked on my laptop, and I'd lock in for hours at a time while he napped.
The conviction of our users gave us the confidence to take the leap. When the product worked, people felt its magic, and we could feel it, too. So even with some remaining technical corners to smooth out, we decided to launch in November. Over the next couple of weeks, we saw that word of mouth was the primary traffic source, and we wondered why we didn’t launch sooner.
But we hardly celebrated. Instead, we were fixated on the same error rates we’d been monitoring all summer. We’d join a Friday All Hands and clap at a big number like $10 million ARR. But none of that provided joy or value. It just validated that we were right to launch and told us to keep going.
If you listened to one of our all-company meetings, you’d think the company is doing terribly. Even when things are going really well, we’re focused on what we could be doing better.
The rebrand: Care in, care out
Throughout all my design work, I think about the difference between something being usable and desirable. I believe that across any medium — fashion, music, food — you can tell when something is made with care. The same was true of my Saturday family dinners as a kid. I don’t think software is any different.
So when I originally created our first logo for the November launch, I gave myself two goals: We need a memorable jewel, and we need to communicate that we care. I wanted someone who first interacts with Lovable to think, “Oh, there’s a proper team behind it.”
I made the first brand hero in Spline with the Google Material Design symbol for the heart icon. I thought this big, glorious Technicolor 3D heart was certainly distinctive, and showed people that we cared.
It met both of my goals, but it broke every rule of scalable design. The favicon was unreadable. The colours all turned to mush. The team desperately wanted merch, and I was horrified by the thought of memorialising my crappy proto-brand on a T-shirt forever.
Two months after the launch, we had enough breathing room to rebrand to something that would scale with our newfound growth. I reached out to Nick Pattison at Primary Studio, whose work I’d long admired. He’d worked with legendary designer Milton Glaser, the creator of the “I Heart NY” logo. Together we had the small goal of coming out of a two-week sprint with a mark that could endure decades.
I put together an extremely simple brief. I went back and rewatched all the calls I’d taken with users over the past month to try to distill the experience of using Lovable. Whenever someone’s body shifted, I wrote down what they were talking about. My thesis was if their body changed, the chemicals in their brain must be changing, too. These calls gave me the language for three brand emotions that we still use to this day:
- Possibility: “I never thought doing this would be possible.”
- Empowerment: “I don’t need to rely on anyone else to do what I want to do.”
- Speed: “I feel like I have superhuman abilities.”
I said to Nick, “If we’re promoting these emotions, we’re on brand. If we’re demoting them, we’re off brand.” The only other requirement I gave him was no purple. Completely subjective, but for some reason I have an irrational aversion to purple.
Over the next few days, he and his collaborator showed me a series of two options to react to. We’d started on a Monday, and by Wednesday, we were playing around with dreamscapes — led by the idea that you feel possibility, empowerment and speed in your dreams. We tried vapor-wavy, almost psychedelic elements.
It was a direction, but not quite a brand yet. So we asked: How can we reduce a dreamscape into something precise? What’s the essence of it? And so the heart was born: turned to its side, like an L (which most people don’t realise). The gradients and texture evoke the haziness of a dream. The spectrum of colours represents the idea that Lovable is a horizontal product, and can be used in unlimited ways. The heart shape ties together our name, and my long-held brand philosophy: Software should make you feel something. Anton had named the company Lovable because he wanted people who created software with our tool to elicit positive emotions in their users.
People have asked me why we made choices that bucked the design zeitgeist in tech at that moment: Most either wanted to look advanced (and therefore technical), or used hand-drawn, irregular motifs in an attempt to feel human. We chose gradients and a heart. We didn’t choose this look to stand out though, even if it helps us do so. It just felt authentic to what people felt using Lovable.
The largest degree of ownership I’ve felt in my career is being the solo steward for the entire brand for the majority of the company’s life. I was trusted to make our first one, and it didn’t scale beyond our initial launch. I recognised the problem and solved it quickly, and brought in the right people to help me solve it. I didn’t need to constantly get the founders’ stamp of approval. I was just trusted to find what felt good to me. Just weeks after the brand sprint, we rolled out the new logo.
The trust to make decisions and live with the consequences
After launching in November 2024, Lovable blazed to $100M ARR just eight months later, becoming one of the few startups to ever do so at that speed. The pace we were growing meant we couldn’t stop the ship to think over every decision; everyone here is trusted to follow their instincts to make the call (don’t bother bringing it to Anton before you do).
Speed can be a dirty word because now it’s associated with slop, or doing something poorly. But my view on speed is that it accelerates the rate of learning, not just the rate of doing. Instead of approaching a problem like an academic and debating it for weeks on end, you can just break it down into smaller parts and give it to real people to solve.
Anton makes it easy for everyone at Lovable to move this way. At every All Hands, he’ll share something he’s been thinking about and say, “Tell me why I’m wrong.” He invites feedback from everyone, all of the time. He’s always looking to uncover his blind spots. That sets the tone for everyone here to realise it’s okay to be wrong. Instead, we’re truth seekers, and it doesn’t matter where the truth comes from, as long as we’re acting quickly to find it.
When I talk to candidates, I try to challenge myself to articulate the values at Lovable that are uniquely Lovable. The one I always come back to is ownership. Everyone here can go from speaking to a user to the first germ of an idea to building it, shipping it, doubling down on it to make it great, and even deciding to delete it. You have full licence to do that here as an IC.
A lot of companies do ownership theatre, where they say they’ll empower you to make your own decisions, but PS, here’s the million gate checks you’ll have to pass, and a VP who’s thought about the problem you’re solving for 30 seconds will helicopter in and change the decision at the last second. Here you’re expected to be the most informed person to make the call, and even when the call is wrong, you’re trusted to find the better path.
You can’t beat someone who’s having fun
Selfishly, as Lovable scales, I'm trying to build the team that I’ve always wanted to work on.
In hiring, I spend a lot of time digging into people’s intrinsic motivations — namely what they did in childhood. The people who are now thriving on my team have a story that’s very similar to mine: They got into design because they were creating Winamp skins or banners for Call of Duty. They were nerdy kids who spent too much time on the internet, and now they care deeply about the designs they make because it feels like an extension of play. Those are the first two designers I hired. Niklas grew up taking computers apart for fun. Felix started his first software business while still in school.
What I’ve learned is that you just can’t beat someone who’s having fun. Everything I now do professionally I learned from my curiosities as a kid, and so I want to find people who are similarly in tune with what makes them tick.