Reference of no code, Lovable claims 50 million projects created. Its co-founder and CTO Fabian Hedin details his strategy to transform the platform into a proactive agent, capable of supporting its users beyond code generation.
JDN. Lovable faces intense competition, both from Vibe Coding tools like Replit or Cursor, but also from large AI laboratories like Anthropic with Claude Design. How do you differentiate yourself? ?
Fabian Hedin. First, by building a great product, because that’s ultimately what users base their choice on. Lovable has become the platform that defines this category, not by being the first to launch in this market, but by being the one that users prefer. 80% of our users come to our platform with the objective of creating an application to develop a business and generate income. We don’t just produce code that then has to be deployed elsewhere. We manage the entire process, from creation to deployment, integrating the building blocks you need, such as the infrastructure and database with Lovable Cloud, or the collection of payments with Lovable Payments. The objective is to facilitate the entire process, from your idea to its realization until its launch and thus enable the end-to-end success of the projects created by our users.
Lovable claims 50 million projects and a million new projects created every week. Who are your users today?
The uses are very varied. We have individuals, freelancers, small structures and more than half of the Fortune 500 companies among our clients. With the power of these foundation models, it is now possible to develop a product that can serve different segments and verticals. Our users therefore do not have to choose a single use case when using our tool. In the old world, you probably would have used twenty different SaaS tools to solve a problem, compared to just one today. Projects built with Lovable accumulate 720 million monthly visits, or 27 times more traffic than the Lovable site itself.
Is your goal to replace all SaaS platform subscriptions with a Lovable subscription? ?
Building your own internal tools is a major use case for Lovable, whether you are a startup or one of the largest companies in the world. Two applications developed on Lovable can thus very easily connect to each other since our agent has the context for each. Integration with third-party tools also becomes simpler. Simply put, if your goal is to create an internal tool for your company, you should be able to build it yourself, and that’s the trend we’re seeing. This gives users more control and allows more context to be added within these tools.
So, is this the end of SaaS as we know it?
Many traditional SaaS providers will remain relevant, but they need to rethink their offering. Do their users want to manage twenty dashboards in twenty different interfaces? Probably not. On the other hand, these players have a lot of aggregated knowledge about their customers within their vertical. Turning these learnings into tools that agents like Lovable can connect to is the direction software development is heading today.
Many predict the end of interfaces in the age of agentic AI. Should we now design software products by first thinking about how an agent will access them?
Exactly. We need to build tools that AI agents can connect to. Users will increasingly go through products like Lovable, a single entry point where an agent responds to all their requests. Some companies that have been building SaaS for years are very good at providing these services but they simply need to evolve how they make them accessible.
Are we still far from seeing non-technical profiles being able to develop an application as sophisticated as Airbnb or Uber with Lovable?
If you go to Lovable, and ask to create a clone of Airbnb or a similar platform, I am sure that you will already obtain a very good result, especially after a few iterations. But I want to emphasize here that growing a large business like the ones mentioned requires much more than software development. That said, building a business like the ones you mention requires much more than software development. We have a considerable number of non-technical users who have started a business and are generating revenue from what they have created on the platform. I can quote Sabrine, a Brazilian user, who created the Plinq platform allowing women to check the background of the person they are meeting with. Or Swedish startup Klar, which developed an AI-powered learning agent on Lovable that reached $13,000 in annual recurring revenue in its first month.
How much of the quality produced by Lovable really comes from the underlying models, and how much comes from the harness, in other words your orchestration layer, the prompts, or even the tool calls?
It is very difficult to achieve the result produced by Lovable by simply using a raw model. This illustrates the value we add on top of these foundation models. We work with all AI labs and continually evaluate models to ensure that whatever your request, the best of them are used behind the scenes. Each prompt sent to Lovable is not a single call to a model, but a larger orchestration where a multitude of small models go through the codebase, identify what is relevant, and feed that context into a larger model which then writes the code. Depending on the task, we use different models, which requires excellent evaluation skills in a constantly changing environment. This is also why we have chosen not to expose this information to our users. They don’t have to worry about it, we make the best decision possible for them.
Is dependence on models developed by other companies a risk? As a CTO, do you plan to eventually train your own models in order to gain independence?
Of course, a company must always examine the risks to its business, and model independence is clearly one for software players. Most companies think about it, and so do we. I can’t detail what we’re doing in this area, but we have a lot of high-quality data and we’re well-positioned to improve our users’ experience at every level, from the orchestration layer all the way down to the model layer.
You announced last June a multi-year partnership with Google multiplying your use of their infrastructure by five. Which part matters more: access to Google Cloud or Gemini models?
This is an important partnership. We produce a lot of software and we must therefore guarantee the computing power necessary to serve our users. On models, as I said, we work with practically all suppliers and our promise is to always provide access to the best model, which varies according to use. But we clearly have good use cases for the Gemini family as well.
Do you plan to simplify the deployment of native mobile applications, particularly in the App Store?
We’d really like to make this smoother, but ultimately Apple largely decides how easy the process will be. The web has been an excellent field for us, and there is still a lot to do with responsive web applications. This openness of the web is also one of the reasons for our growth, since it allows us to publish a project and share it with anyone. That said, we are exploring the possibility of creating full native versions, and we would like to simplify publishing to stores in the future.
You launched General Tasks in March, allowing you to generate pitch decks, marketing videos or reports. Is the goal to make Lovable a general productivity platform, beyond creating applications?
We have achieved something great in the field of creating websites and web applications and we definitely don’t want to lose that. This is our gateway to a wider market and we must continue to invest in this area. That said, when we look at how people use our tool, we see that they want to do more. After creating an app, the logical next step is to need to present this product to your audience, therefore to create a pitch deck, or to understand the behavior of your users, therefore to analyze data, etc. Our goal is to make Lovable the co-founder of your start-up, not just a developer agent.
The success of tools like OpenClaw or Hermes shows the appetite for agents capable of carrying out tasks ever more autonomously. What are your initiatives in this area?
Indeed, being able to do more things with an agent seems to be a natural evolution for our platform, and this is also the meaning of General Tasks. The trend, and what we’re working on, is to make Lovable more proactive. By combining the agent’s ability to write code with integrations to your current tools, we are moving towards a platform that offers more productivity, and this is the part we want to push even more. Today you ask Lovable to do things for you. The next natural step will be that Lovable can directly recommend certain actions by saying, “I have prepared a first draft, what do you think?”. Maybe he’ll even test some of these ideas with your users for you.
According to SiftedLovable is reportedly in discussions to raise $300 million at a valuation of $13.2 billion. Your co-founder has also said he is interested in acquisitions. What’s next for Lovable?
Regarding fundraising, we have nothing to announce. Last May, we claimed $500 million in annual recurring revenue. Lovable has 250 employees and should have 400 by the end of the year. On acquisitions, we have many former start-up founders in our ranks and we like to work with people who have a lot of ambition and autonomy. It turns out that many of these profiles now run their own companies. Buying companies is therefore a great way to attract these people that we would love to work with. Looking forward, our goal has always been to unleash human creativity and democratize access to software development. Creating sites and applications must become even simpler, to allow everyone to launch ever more ambitious projects. And then we want to help you grow them.
Fabian Hedin is the co-founder and CTO of Lovable, a software creation platform that allows everyone to turn their ideas into applications. Before Lovable, he was founder and CEO of Tentium, then founder and CTO of Tenfast. He also held the positions of Frontend Manager at Depict.ai and Fullstack Engineer at CurbFood. He lives in Stockholm, Sweden, where Lovable is headquartered.
