Lovable FAQ: The Real Questions People Ask Before They Build

I get the same handful of questions about Lovable almost every week. From people who just heard about it, from founders weighing it against Bolt or Cursor, from developers who want to know what is actually under the hood. This post is the answer key. No marketing, no fluff, just the questions people ask and what I tell them.

If you are brand new, start at the top. If you already shipped something on Lovable and you are here for the pricing or the export question, jump to the section you need from the list below.

Getting started and the first prompt
Pricing and credits
The tech stack and the model behind it
Workflow, Plan mode, and the Knowledge Base
Output, ownership, and exports
Common pitfalls and how to avoid them

Getting started

Can I really launch a working app just by describing it?

Yes, for a wide range of apps. Landing pages, dashboards, internal tools, marketing sites, CRUD apps backed by a real database, AI chat interfaces, booking forms, simple marketplaces. Lovable writes the React code, wires the backend, and gives you a live preview on every prompt. You see the app build itself as you type.

The honest caveat. Complex products with heavy custom business logic, real time multiplayer, native mobile, or anything that needs a specialist engineering team are not a one prompt job. You can still start them on Lovable, but expect to iterate, review the code, and bring in a developer for the hard parts. For a full walkthrough of what a first build looks like, read how to build an app with AI.

What is the difference between Plan mode and Build mode?

Plan mode is for thinking. The agent reads your project, asks clarifying questions, and proposes a plan without changing any files. Use it when you are about to make a bigger decision, like restructuring routes, adding auth, or refactoring how data flows. You approve the plan, then switch to Build.

Build mode is for shipping. The agent edits files, runs commands, and updates the preview. Most short tasks live here. The rule of thumb. If you could explain the change in one sentence, just Build. If it touches five files or you are not sure how it should work, Plan first. Planning costs fewer credits than fixing a wrong build.

Should I start with a template or from scratch?

From scratch, unless your idea maps cleanly onto a template. A blank prompt forces you to describe what you actually want, which usually produces a tighter result than a template you have to bend. Templates are great when the shape is obvious, like a SaaS landing page, a portfolio, or a basic dashboard. For everything else, write a clear prompt and let Lovable build the skeleton from your description.

What is the Knowledge Base and how do I use it?

The Knowledge Base is a small document attached to your project that the agent reads on every prompt. Put your product context there. The audience, the tone, the brand colors, your tech preferences, the things you never want changed. A good Knowledge Base saves credits because you stop repeating yourself in every prompt.

Keep it short. Half a page is plenty. Long Knowledge Bases dilute the signal and make every response slower. Update it when your product direction changes, not every week.

Pricing and credits

How does the credit system work, exactly?

A credit is roughly one round trip with the AI. Ask for a new component, request a refactor, fix a bug, wire a database table. Each of those is one credit if the change is simple, more if the agent has to touch many files or run agentic loops. Editing code yourself in the inline editor is free. Visual edits like changing text or colors in the side panel are free. Previewing and publishing are free. You only spend credits when the AI does real work.

What are the current plan prices?

Free is zero dollars with five daily credits, capped around thirty per month, public projects only. Pro is twenty five dollars per month with one hundred credits, private projects, custom domains, GitHub sync, and full Supabase access. Teams starts at thirty dollars per seat with pooled credits, shared workspaces, and centralized billing. Enterprise is custom pricing with SSO, audit logs, and volume credit packs. Annual billing usually saves about twenty percent. Full breakdown lives in Lovable pricing explained.

Which actions cost credits and which are free?

Free actions. Inline code edits, visual edits, publishing, connecting a custom domain, browsing project history, opening previews, downloading code, and using the file tree. Paid actions. Every prompt to the AI agent, agent loops that span multiple files, image generation, and AI driven refactors.

How do I avoid burning credits on debugging loops?

Three habits make the biggest difference. First, never let the agent loop on the same error more than twice. Stop, read the error yourself, paste the exact log into a new prompt with the file path, and tell it what you suspect. Second, prefer Plan mode for anything ambiguous, then Build the agreed plan. Third, batch your asks. Five small prompts cost five credits. One clear prompt that asks for the same five changes costs one or two.

Do monthly credits roll over?

No. Credits reset each billing period on Pro. If you do not use them, they are gone. The exception is Teams, where credits pool across seats so a quiet seat does not waste budget. Plan your bigger builds for the start of the month so you have room to iterate, not the last week when you are out.

How do I actually cancel my subscription?

Open Settings, then Billing, then Manage Subscription. Cancel from there. You keep paid features until the end of the current period and then drop back to Free. Your projects stay yours either way. The official steps live in the docs and the link is in your billing page.

Tech and stack

What AI model powers Lovable?

Lovable runs on Anthropic Claude as its default coding model, with newer models slotted in as they ship. You do not pick the model manually. Lovable routes the request, handles the system prompts, and adds project context so the model sees what your codebase looks like. The benefit. You get a state of the art coding model without managing keys, rate limits, or prompts.

What is the default tech stack?

React with TypeScript, Tailwind CSS, TanStack Router and Start for routing and server functions, and Supabase for the database, auth, storage, and edge functions when you turn on Lovable Cloud. Vite is the build tool. The output is ordinary modern web code that any React developer can read.

What is Lovable Cloud versus connecting my own Supabase?

Lovable Cloud is a managed Supabase backend Lovable provisions for you. No dashboard, no keys, no migrations to babysit. Auth, database, and edge functions just work. Best for solo builders who do not want backend overhead.

Bring your own Supabase if you already have a project, your team needs direct dashboard access, or you have compliance requirements. You paste your URL and keys, Lovable uses them, and you own everything in your own Supabase org.

Can I import an existing project from GitHub?

You can sync any Lovable project to GitHub in one click and push or pull from there. Direct import of an arbitrary existing repo into Lovable is limited. The agent works best on projects that started in Lovable and follow its conventions. For an outside repo, the practical path is to create a new Lovable project, push your existing components and styles into it gradually, and let the agent learn the structure as you go.

Building and workflow

How fast can it realistically ship an app?

A simple landing page with a contact form, fifteen minutes. A small SaaS dashboard with auth and a database, an afternoon. A real product with payments, multiple roles, and polished design, a long weekend. Anything beyond that, plan for weeks of iteration like any real software project. Speed comes from clear prompts, not from skipping the thinking.

What is the Visual Editor and when should I use it?

The Visual Editor lets you click any element in the preview and change its text, color, spacing, or class without writing a prompt. It is free, instant, and perfect for copy edits, color tweaks, and small layout fixes. Use it for anything that does not need new logic. Save the prompts for changes that touch behavior, data, or new components.

How many prompts does a typical project take?

A landing page lands in five to ten prompts. A small CRUD app in twenty to forty. A real product with auth, payments, and admin in one hundred plus, spread over weeks. The exact number depends on how specific your prompts are and how much you use Plan mode and the Knowledge Base.

What should I do before writing my first prompt?

Write two paragraphs describing the app. What it does, who it is for, what the first screen looks like, what data it stores. Then list three to five core features. Then paste that into your first prompt as the starting brief. You will save hours of back and forth and your first build will be ninety percent of the way to what you actually want.

What is a PRD and should I write one first?

A PRD is a product requirements document. For a Lovable project, you do not need a formal PRD. You need a short brief. The two paragraphs from the question above, the list of features, a sentence about the audience, and any non negotiables like brand colors or a specific tech choice. That short brief, dropped into the Knowledge Base, is your PRD.

Output and ownership

Can I export the code?

Yes. Push to GitHub from Settings, then clone the repo and run it anywhere. The output is a standard React, TypeScript, Vite project. You can keep building in Lovable, in your own editor, or both. Nothing about the code is locked to the platform.

What happened to GPT Engineer, is it the same thing?

Lovable is the team and product that grew out of GPT Engineer. The open source GPT Engineer project still exists for the developers who liked the CLI. Lovable is the hosted product with the visual builder, the agent, and the deploy pipeline. Same company, different product surface.

Can it produce production ready apps or just prototypes?

Both, depending on how you treat it. Many shipped products in 2026 are running on Lovable code as the primary stack. The discipline is the same as any other codebase. Review what the AI writes, set up auth and RLS properly, test the critical paths, and do not ship a credit card form you have not read line by line. For real apps people have launched, see does vibe coding work.

Do I still need to understand code to use it effectively?

For a landing page or a personal tool, no. For a real product, a little. You do not need to write React from scratch, but you should be able to read it. Understanding what a component is, what props are, how state changes, and how an API call works will save you days of frustration when something breaks. Lovable shortens the path from idea to shipped, but it does not remove the responsibility of knowing what you shipped.

Common pitfalls

The same mistakes show up in almost every project. Vague first prompts that try to describe an entire app in two sentences. Letting the agent debug the same error six times in a row instead of stopping to read the log. Ignoring the Knowledge Base, then wondering why the tone of every response is off. Adding ten features at once instead of one, then losing track of what broke.

Fix every one of these by slowing down for thirty seconds before you press send. A good prompt saves five bad ones. A clear Knowledge Base saves a hundred. If you want the full list with examples, the beginner mistakes guide walks through each one.

FAQ that did not fit a section

Is there a free trial of Pro?

No formal trial, but Free is generous enough to evaluate the whole product. Build one tiny app, ship it, decide.

Can I use my own domain?

Yes on Pro and above. Buy the domain anywhere, point the DNS to Lovable as the docs show, and connect it from the project settings.

Can Lovable build mobile apps?

Lovable ships web apps. Responsive design works out of the box, so most products work great on a phone browser. For real native iOS or Android, export and use Capacitor or React Native around the shared logic.

Does Lovable support multiplayer or real time features?

Yes through Supabase real time. The agent can wire subscriptions and presence, but for heavy multiplayer use cases you will spend more time guiding the design than on a simple CRUD app. Plan first, then build.

How does Lovable compare to Bolt, Cursor, or Replit Agent?

Different shapes. Lovable is the most visual, the most opinionated, and the easiest to publish from. Cursor is the closest to a traditional IDE with AI inside it. Bolt sits in the middle. Replit Agent is closer to a hosted dev environment with agent help. Full breakdowns live in Lovable vs Bolt.new and Claude Code vs Cursor vs Copilot.