Lovable

What Is Lovable?

Lovable takes a description of a product and builds it — not a mockup, not a wireframe, but a working web application with a front end, a database behind it, user accounts if it needs them, and a live URL you can send to someone. You describe the change you want next, and it makes that too.

The category this belongs to acquired the name «vibe coding», which undersells it. What is actually happening is that the gap between having an idea and having something people can click has collapsed from weeks to an afternoon. Whether the result survives contact with real users is a separate question, and an important one.

Crucially, and unusually, you own what comes out. Lovable states plainly that the code, the apps, the customer data and the AI output are yours, and projects sync to GitHub. That single fact is what separates this from the generation of app builders that held your work hostage.

How It Works

Plan First, Then Build

There are two modes, and the distinction matters both for results and for cost. Plan mode is a conversation about what you are making — cheap, and the right place to argue with the model about structure before it writes anything. Build mode does the work: generating screens, wiring data, deploying.

Using plan mode properly is the single biggest difference between people who get something usable and people who spend a weekend fighting a half-built app. Decide the shape first; the model is much better at executing a clear plan than at inferring one.

A Real Stack, Not a Sandbox

Applications come with hosting, a database, authentication and file storage as part of the platform, so a prototype does not need a separate afternoon of infrastructure work before anyone can use it. You can publish on a Lovable subdomain immediately or point a custom domain at it on a paid plan.

Because everything syncs to GitHub, the escape hatch is always open. When the AI reaches the edge of what it can do — and on anything non-trivial it will — a developer can take the repository and continue by hand. That is the difference between a prototype and a dead end.

Connections and Design Systems

Connectors link the app to outside services, an MCP server exposes Lovable to AI assistants you already use, and design systems on paid plans keep generated interfaces consistent with your brand rather than with the model's default taste. Design templates and shared systems are what make this usable by a team instead of one enthusiast.

Credits: How You Actually Pay

Lovable bills by credit, and the details are worth ten minutes before you subscribe, because several different kinds of credit behave differently.

A message in plan mode costs one credit. A build action costs between roughly half a credit and two, depending on how much it has to do. Running the app afterwards consumes separately, scaling with database size, storage, traffic and compute — and AI features inside your app are metered by model and token use on top of that.

Expiry rules differ by type. Monthly plan credits roll over while your subscription stays active but expire about two months after they are issued. Purchased top-up credits last a year. Free daily build credits expire at the end of the day, and the monthly cloud and AI grants do not roll over at all. The pattern is consistent: nothing accumulates indefinitely, so buying capacity ahead of need is wasted money.

What It Costs

Annual billing gives two months free. Students can get up to half off Pro with verification, and there are separate arrangements for universities and for schools.

Note the thing the table makes obvious: Business costs twice Pro and includes exactly the same hundred credits. You are buying governance — SSO, access control, a security centre, a shared workspace — not capacity. If it is just you, or a small team that trusts each other, Pro plus top-ups is the cheaper answer.

The free tier is real but small: five build credits a day is enough to see whether the idea works and not enough to finish anything. Budget for top-ups on top of whichever plan you choose, because the running cost of a live app is separate from the cost of building it.

Who Gets the Most From It

Founders Testing an Idea

The strongest case. A landing page with a working sign-up, a functioning demo for an investor meeting, a first version that ten people can try — all of it before hiring anyone. The value is not the code; it is finding out in a week whether anybody wants the thing.

Product Managers and Designers

Showing beats describing. A clickable, data-backed prototype settles arguments that a Figma file leaves open, and it takes about the same time to produce once you are used to plan mode.

Teams Building Internal Tools

The dashboards and admin panels that sit at the bottom of every engineering backlog forever. Nobody will ever prioritise them, and they do not need to be beautiful — they need to exist.

Developers Skipping the Boring Part

Scaffolding, auth, CRUD screens, deployment. Generate the skeleton, pull the repository, and spend your attention on the part that actually required you.

What to Watch Out For

• Credit costs are not intuitive: building, running the app and in-app AI features are metered separately.
• Credits expire — monthly ones roughly two months after issue, daily free ones the same day, cloud and AI grants not at all.
• Business costs double Pro for the same hundred credits. You are paying for governance, not capacity.
• Generated code is workable, not idiomatic. A developer inheriting it will want to refactor before building on it seriously.
• The further you get from common patterns, the harder the model works and the more you pay for diminishing results.
• Security is your responsibility. AI-generated apps have shipped exposed data and missing access checks; review anything that touches real user information.
• Running costs scale with usage, so a prototype that succeeds gets more expensive precisely when you are least ready for it.
• Free daily credits reset every day, which makes the free tier a trial rather than a way to finish a project slowly.

Frequently Asked Questions

Is Lovable free?

There is a free plan with five build credits a day, up to thirty a month, plus small monthly allowances for cloud and AI usage. No card is needed. It is enough to evaluate the product and not enough to complete a real project.

Do I own the code?

Yes. Lovable states that you own the apps and sites you build, the customer data stored there and the AI output, subject to third-party rights in the underlying models. Projects sync to GitHub, so you can take the repository and leave at any point.

What is a credit?

The unit Lovable meters usage in. A plan-mode message costs one; a build action costs from about half a credit to two depending on complexity. Hosting your app and running AI features inside it consume separately, based on resources used.

Do credits roll over?

Monthly plan credits do while your subscription is active, but expire around two months after being issued. Top-up credits last a year. Daily free credits expire the same day, and monthly cloud and AI grants do not roll over.

Can I use it for a production app?

For internal tools and small products, plenty of people do. For anything holding sensitive customer data, have someone review the access controls and data handling before launch — the model writes plausible code, not audited code.

Why is Business twice the price of Pro?

Because it sells control rather than capacity: shared team workspace, role-based access, SSO, a security centre and priority support, with the same monthly credit allowance. Buy it when the organisation needs the governance, not when you need more building power.

The Bottom Line

Lovable is the clearest demonstration that building a working application is no longer gated on knowing how to build one. For prototypes, internal tools and the first version of a product, it compresses weeks into an afternoon, and because you own the code and it sits in GitHub, that afternoon is not wasted if the project outgrows the tool.

Spend time in plan mode before you build, expect to pay for running the app separately from making it, and get a human to look at anything that stores other people's data. Used that way, it is one of the highest-leverage tools available right now.