AI Giants: Inside Lovable's $100M ARR Journey

Eight months. 

That's all it took for Lovable to reach $100M in annual recurring revenue - making them arguably the fastest-growing software company in recorded history.

In our second episode of the AI Giants podcast series, Codacy CEO, Jaime Jorge, sat down with Mindaugas Petrutis from Lovable to understand how a 50-person European team achieved what most companies never accomplish.

The conversation revealed surprising insights about democratizing software development, the unexpected problems being solved, and why "what's a prompt?" might be the most important question in tech.

Watch the full recording here

If you’re not familiar with Lovable, here’s a quick glance:

Lovable is an AI platform that transforms natural language descriptions into full-stack web applications. 

Users can describe what they want to build - or with their new Voice Mode, literally speak their ideas - and Lovable generates the complete application including frontend, backend, and database. No coding knowledge required. 

The platform has attracted over 2.3 million users, with 180,000 paying subscribers driving their unprecedented growth to $100M ARR in just 8 months.

TL&DR: What Companies Can Learn from Lovable's Journey

Lovable reached $100M ARR in 8 months by solving a simple problem: making software development accessible to everyone. 

Their approach offers critical lessons for enterprises adopting AI development tools. 

• Small teams (50 people) can outpace giants through ruthless prioritization. 
• Voice interfaces matter more than UI - they determine who can actually use your tools. 
• The real opportunity isn't in venture-scale problems but in the millions of 10,000-person problems that become solvable when non-technical experts can build.
• Security can't be an afterthought when your builders don't know what an API key is.
• And perhaps most importantly: 80% of Lovable's revenue comes from complex applications, not prototypes 

From Desperate Founder to Lovable's Growth Lead

Mindaugas Petrutis's path to Lovable wasn't typical. He was a non-technical founder who found himself without an engineer when his product needed building. Like many founders in this position, he turned to AI coding tools out of desperation - and discovered Lovable.

"I was a super user already before joining Lovable. I was obsessed with the product. I was building with it," Mindaugas explained. "I had a product idea last year that I was working on with a small team, but then found myself without an engineer and all of these tools kind of had been popping up already."

What started as a necessity became an obsession with Lovable as a tool. "I set myself a goal like 30 minutes every day to just make something. Honestly, nothing that anybody needs, but it was more to test the capabilities, the limitations, and just see kind of what's possible, really."

He built over 100 projects on Lovable - sometimes without even having an idea: "Sometimes I would even not have an idea on a morning. And I would just say to ChatGPT like, Hey, I have access to this AI coding assistant. I have no idea what to build. Can you just give me some, like, five random ideas? And I pick one of them and build it."

The breakthrough came when he used Lovable to solve a problem he deeply understood. "I went from idea on a Sunday night to first paying customer four days later on a Thursday."

The next morning crystallized everything: "The next day, I built the stripe integration so people could pay for it, like while drinking coffee over breakfast. And I've worked, you know, as part of teams and companies where you're building the Stripe integration, and that would require engineering time previously. And I just did it myself. And it worked."

"I remember just pushing my chair back and being like: ‘Okay, this is insane’."

That product he built with Lovable continues generating revenue today with zero promotion: "I don't promote it. But I think it's done, I think $7000-$8,000 in revenue at this point, people find it, they pay for it. My stripe pings like a few times a week, you know, $50, $50, $50."

When Lovable reached out to recruit him after seeing his success as a power user, the decision was immediate: "When they reached out and came calling,I literally dropped everything. And it was the easiest yes of my career to jump on board."

How a 50-Person Team Built the Fastest-Growing Company Ever

Until very recently, Lovable operated with just 50 employees while achieving unprecedented growth. The secret lies in their almost painful approach to prioritization.

"Being on the inside, you're being thrown every kind of collab integration partnership, whatever you can think of," Mindaugas explained. "What I saw was just a very strict focus on priorities."

The company's discipline is best illustrated by their Slack channel called "Crazy Ideas." When a user requested a feature that Mindaugas thought made sense, he posted it there. Anton, one of Lovable's founders, responded immediately:

"First thing he wrote was, 'I have wanted this for such a long time' and instantly followed up with, 'but it's not priority, we're not doing it.'"

This wasn't an isolated incident. As Mindaugas noted: "If the founders are saying 'I think we should have this, but it's not a priority, therefore we're not doing this' - there's clearly a very well thought-out focus."

When Lovable raised $200 million at a $1.8 billion valuation - becoming a unicorn - their celebration was deliberately modest. While Mindaugas noted that some Silicon Valley companies might "spend half of that funding money on the announcement and on the party," Lovable took a different approach.

"I walked into that party and thought this could have been just a cool person's birthday party. It was super chill - maybe 100 people, we had cake, some drinks, and a unicorn balloon. That was really it."

The understated celebration reflected the company's broader philosophy: "It showed how we're thinking about ourselves as a company, as a team, what the priorities are."

This focus on building over celebrating has enabled their small team to move faster than companies many times their size. As Mindaugas observed from his experience at other companies: "I've been part of companies and teams where [changing priorities] completely derails everything, reshapes direction overnight, and that is never a good story."

Voice Mode, the Death of the Prompt, and the Last Piece of Software

Lovable recently launched Voice Mode, and Mindaugas sees it as crucial for mass adoption - but not for the reasons you'd expect.

"We all live in a tiny bubble called tech," he said. "The rest of the world doesn't operate in that environment. There are hundreds of millions, billions of people that don't know what a prompt is."

Voice changes everything. Instead of learning prompt engineering, users can just talk through their problems. Mindaugas discovered this himself: "I find AI works best when you give it lots of context. To give lots of context by typing, I start skipping stuff. With voice mode, I can just brain dump as much context as possible - here's the idea I want to build, here's what I know about the users, here's what other solutions exist."

Voice Mode aligns perfectly with Lovable’s tagline: "the last piece of software" - a phrase that sounds like marketing fluff until Mindaugas explains what they actually mean.

"It means whenever you want to build something, that's the thing you're going to use. No other tools will be involved. You describe your idea from idea to full working product and it's built. That's it."

They're backing this up by systematically adding every tool a business needs. Analytics just launched. SEO is coming. 

The goal is to support the entire journey, "not just from idea to product, but from idea to a company," as Mindaugas put it. "There's a lot more involved in building a company than just building a product - you need to know how to sell, market, incorporate."

80% of Lovable’s Revenue Isn't Coming From Toy Apps

You’d assume that Lovable makes its money from hobbyists. But the reality at Lovable is that 80% of their revenue comes from complex applications - and the stories behind them are wild.

Take the wealth management platform disaster. A non-technical founder with multiple companies had been burning money for two years trying to build this product with a team of engineers. Two years. Nothing to show for it.

"The person then discovers Lovable, builds a prototype, shows to his team and his team is like, can we finish it in Lovable?" Mindaugas recounted. "They built it within a week and the first client was onboarded within one week."

But the more interesting story might be the guy in rural America that Mindaugas talked to last week. Mid-50s, works at a disability services agency, had never really used ChatGPT. Through a music platform partnership, he stumbled onto Lovable.

This guy discovered something nobody else was seeing: "A specific problem for around 10,000 people in the state where he lives that nobody's solving."

Why had nobody solved it? The math doesn't work for traditional development. "Insurance companies are not going to touch it - it's 10,000 people. It's just not worth it. His company is small. They're not going to hire a bunch of engineers."

With Lovable, he built a prototype, showed his CEO, and got the budget to develop it further.

This is the real disruption. Not the unicorn valuations or the venture-scale opportunities, but the millions of 10,000-person problems that suddenly become solvable. As Mindaugas put it: "There are millions of these weird in-between problems... too small for larger companies, not large enough for startups."

The person solving them isn't a developer or a funded startup. It's someone who actually experiences the problem daily and now has the tools to fix it.

Lovable ran another experiment that revealed just how many problems are waiting to be solved. They partnered with Outside Lands music festival in San Francisco for a hackathon - but unlike traditional hackathons that require coding skills, this one was open to everyone. Musicians, DJs, producers, talent managers - people who live and breathe the festival experience but can't code - suddenly could build solutions.

They built apps for the festival experience and the music industry. The top 10 were integrated into the official festival app for attendees to test and vote on. But here's what matters: weeks later, festival organizers told Mindaugas that two or three of these apps were so useful they want them officially integrated into next year's festival.

Think about that dynamic. The festival would never identify these problems themselves, and even if they did, they wouldn't hire engineers to solve them. But festival-goers who experience these problems? Give them Lovable and they solve them in a weekend. As Mindaugas put it: "The festival itself will never be as aware as their attendees of what those problems are."

The Security Challenge When Everyone Can Ship Code

When non-technical people ship production code at scale, security becomes an entirely different problem. Lovable is grappling with this reality daily.

"A couple of cases like a few months ago," Mindaugas recalled, "people were putting in API keys and then that was being built into the front end and then that was being exposed or people were grabbing them."

Lovable now automatically flags API keys and blocks them from being exposed. But the deeper challenge is educational. As Mindaugas pointed out: "If a lot of users are arriving to the product and they're like, I don't even know what a prompt is - well, what the hell do they know about security, right? What do they know about backends?"

The solution requires rethinking security from the ground up. "To me, it's an educational product on top of allowing you to build a product. How do you balance the educational piece where you're giving enough context to the user? You're prompting them to make sure they check certain things, but also enabling them to understand."

Lovable is building out their trust and safety team, with Igor Andriushchenko recently joining to lead efforts around "scammers, malicious actors, all of that." 

But the fundamental challenge remains: "How do you design the product to work in such a way where anyone can be confident in making sure that what they're releasing is secure and our AI does enough of the work to ensure that, but then the user also has that confidence and comfort too?"

P.S. This is exactly why Codacy built Guardrails - to provide automated security checks for AI-generated code. As more non-developers ship production applications, the traditional assumptions about security education no longer apply. The code needs to be secure by default, not secure because the developer knows to make it secure.

What Happens When Everyone Can Build?

When Jaime asked whether Lovable will create or erase more jobs, Mindaugas was direct: "Definitely create."

His logic: AI will automate many roles, but those displaced workers need somewhere to go. "Those humans have great ideas, but the people being replaced are the non-technical ones. So we're giving them something to start a side project that generates income, start a company that they couldn't before."

The evidence is already visible in Lovable's user base. The disability services worker solving a 10,000-person problem. Festival attendees building apps that get officially adopted. People generating $7-8K monthly from side projects built over breakfast.

"There are millions of these weird in-between problems," Mindaugas explained. Problems too small for enterprises, not venture-scale enough for startups, but perfect for individuals who actually experience them.

Lessons for Enterprise Teams Adopting AI Development

The conversation revealed several practical insights for companies looking to adopt AI-powered development workflows.

Start with the problems nobody wants to solve

Mindaugas's experience shows that AI tools excel at the "weird in-between problems" - too small for dedicated engineering resources but important to specific user groups. These make perfect pilot projects because failure won't derail core operations, but success creates immediate value.

Voice changes adoption rates

"There are billions of people that don't know what a prompt is," Mindaugas noted. For enterprise teams, this means voice interfaces aren't a nice-to-have - they're essential for getting non-technical stakeholders to actually use AI tools. 

His own workflow improved dramatically when he could "brain dump as much context as possible" through voice rather than typing.

Focus >>> everything else

Lovable's 50-person team outpaced much larger competitors through brutal prioritization. Even features the founders personally wanted got killed if they weren't a priority.

For enterprises, this suggests starting with a narrow use case and saying no to expansion until that's perfected.

Security requires new thinking

When users don't know what an API key is, traditional security training fails. Lovable learned this the hard way when users were exposing keys in frontends. 

The solution: automatic detection and blocking combined with in-context education. Enterprises need to assume their AI users know nothing about security and build accordingly.

The AI Development Era is Here and Codacy Can Help

The democratization of software development is happening now. Non-technical users are shipping production applications. AI-generated code is processing real payments and handling sensitive data. Traditional approaches to security and code quality no longer apply when the person deploying code might not know what an API key is.

For teams navigating this transition, automated safeguards become essential. Learn how Codacy Guardrails can help protect your AI-assisted development workflows.