Create an app with Bolt: complete guide and tips
Bolt.new generates a web application in the browser and lets you export the code. It does not compile iOS or Android binaries: store publication happens outside the tool.
Article summary
- Bolt.new, published by StackBlitz, runs a development environment in the browser. Nothing to install on the machine.
- The usual result is a web application in React and Node.js, not a React Native app you can publish as is.
- Paid plans start around $20/month. A precise initial prompt consumes fewer tokens than a series of corrections.
- The code exports in full. Since September 2025, Bolt creates a database per project, which can be migrated to Supabase.
- An Expo integration lets you prepare React Native, but iOS and Android compilation goes through EAS, outside Bolt.
Summarize this article with
Bolt.new passed one million active users in under six months after launch. Founders, designers, consultants, people with no development background, started shipping full-stack prototypes in a few hours. The buzz is real, and the results are too.
What most tutorials leave out is exactly what Bolt produces, and what it will never produce. firstapp hears every week from founders who built a solid Bolt prototype, and who now wonder why their users ask them "so when is the app?".
This guide explains how to create an application with Bolt from A to Z, with the real limits stated clearly from the start.
Newsletter
Once a week.
Last issue
We opened 10 apps to see how they sell a subscription.
+534 people read us each week
What exactly is Bolt.new, and who is it for?
A complete development environment in the browser
Bolt.new is built by StackBlitz, a team known in the web development world. What makes Bolt unique among vibe-coding tools is its architecture: it runs a complete development environment directly in your browser - terminal, package manager, local server, and the rest.
In concrete terms, you install nothing on your machine. You describe your application, and Bolt generates a working codebase you can interact with live. It is a step above a simple interface generator.
Bolt is aimed at a slightly more technical profile than Lovable. You do not need to be a developer, but you should be comfortable with notions such as "frontend", "backend", "API", or "database". Pure non-technical people can use it, but the learning curve is steeper.
What Bolt actually produces
Let us be direct now. Bolt generates web applications, typically in React with a Node.js backend. These applications run in a browser, on desktop and on mobile. They can be deployed on a domain and shared with real users.
What Bolt does not do: generate a React Native application you can publish on Apple's App Store or the Google Play Store. A web application and a native mobile application are two different products - technically, commercially, and in the way your users perceive them.
To remember: Bolt is one of the best tools for prototyping quickly and validating a concept. It is not a native mobile development tool. If your final goal is a presence on the stores, Bolt is a step, not a destination.

What to prepare before you open Bolt
Define your application before you open the tool
Bolt is fast. Too fast for founders who have not yet clarified their idea. The tool executes what you describe with formidable efficiency, but if the description is vague, the result is too.
Before you sign in, answer these questions in writing:
- What problem does your application solve, and for whom exactly?
- What is the main feature, the one without which the application is useless?
- What are the 3 to 5 secondary features, in priority order?
- Do you have technical or regulatory constraints to include?
If you are unsure about your concept, read our guide on how to validate your app idea before building first. Building fast on a bad idea is losing time, not gaining it.
Understand Bolt's token system
Bolt runs on a token system. Every generation, every change, every bug fix consumes tokens. The free plan includes a limited volume, enough to explore, not enough to build a complete MVP.
Paid plans start around $20/month. The essential rule: a precise, complete initial prompt consumes far fewer tokens than a series of successive tweaks to fix an approximate result. Invest time in your starting prompt. It is the best return on investment available with Bolt.
Create your application with Bolt: the key steps
Write a complete starting prompt
Unlike some AI tools, Bolt rewards exhaustive prompts. Where other tools can be thrown off by too much detail, Bolt generally integrates that detail very well from the start.
A good prompt for Bolt should contain:
- The complete functional description of the application (not the design, the functions)
- The type of users and their technical level
- The preferred stack if you have one (otherwise, let Bolt choose)
- The third-party integrations you need (Stripe, Supabase, external APIs)
- Security or compliance constraints, if they apply
An effective example: "Create a web application for client management for freelancers. Features: a client list with a detailed record, project tracking per client, simple invoicing with PDF generation, a dashboard with monthly revenue. Node.js backend, PostgreSQL database, interface in French, professional and sober style."
Iterate feature by feature
Once the first version is generated, Bolt uses a "diffs" system. It rewrites only the parts of the code that need to change. That is what makes it faster than most of its competitors on successive edits.
The recommended method: handle one feature at a time. Generate, test in the built-in preview, confirm everything works, then move to the next. If something breaks, go back immediately. Bolt offers a version history.
Avoid asking for several changes in a single prompt. "Add feature X, change the header style, fix the bug on page Y, and integrate API Z" almost always produces a partial or unstable result.
Export your code
This is one of Bolt's distinctive advantages: it allows a full export of the generated code. You can take your entire project, front and back, and deploy it in any environment, or hand it to a developer to evolve it.
Lovable locks you in more on its platform. Bolt gives you the keys. If you expect a move to native development or a handover to a technical team, code export is a real asset.
Taking back control of the code Bolt generated
Adapt the interface and the design
Bolt generates functional interfaces, but their visual quality is often less finished than what Lovable produces. The components are there, the logic holds, but the rendering frequently needs adjustments to be presentable to demanding users.
You can ask for design changes directly in natural language. Bolt applies them with decent precision for simple changes (colors, spacing, typography). For complex interface changes, you will get better results by going through the exported code and working in an editor such as VS Code.
Change the business logic
This is where Bolt shines compared with its competitors. The application logic - the business rules, the calculations, the conditional flows - can be described and changed through prompts with remarkable precision. If your application has specific business rules, Bolt integrates them better than most tools on the market.
Connect APIs and external services to your app
Database and authentication
Since September 2025, Bolt creates its own databases by default for each new project. You can use them directly or migrate them to your own Supabase account. Bolt supports that migration natively.
User authentication is generated automatically if you mention it in your prompt. Sign-up, sign-in, session management: the result works and is enough for an MVP.
Payments and monetization
Bolt can integrate Stripe to handle one-time payments and subscriptions. The technical setup holds up. But as with any AI tool, the technical part is not the problem. The monetization strategy is.
The moment you present your offer in the user journey, the way you structure your plans, the prices you choose: these decisions have a direct impact on your conversion rate. According to RevenueCat 2026 data, applications that optimize their paywall convert on average 3 to 4 times more than those that show a default offer. No prompt will give you that optimization. It is domain expertise. Our guide on how to generate your first revenue at an app's launch will give you the basics to start on the right foot.
The technical limits to know
Bolt shows its limits on features that fall outside common web patterns. Complex real-time logic, recommendation algorithms, multi-tenant architectures, integrations with undocumented proprietary APIs: these topics generate a lot of iterations and unstable results.
Warning: if your main feature is complex or uncommon, test it first with Bolt before you build the rest. That is where projects get stuck, and where tokens evaporate.
Test your prototype with real users
The onboarding Bolt will not think through for you
Bolt builds what you describe. It does not ask itself whether your user journey is logical, whether the value is felt before sign-up, whether the friction at the entrance is justified. Those are questions you have to ask yourself, and the AI will not answer them for you.
Applications generated by Bolt often repeat the same mistakes: sign-up before any demonstration of value, an onboarding flow that is too long, no identifiable "aha" moment in the first screens. These problems are not bugs. They are design decisions nobody made.
To avoid building onboarding that leaks, read our guide on mobile onboarding optimization. The principles apply directly to a web app.
Test with real users as fast as possible
The value of a Bolt prototype is not measured by the quality of the generated code. It is measured by what your real users do with it.
Share your prototype with strangers, not with people close to you, who will tell you it is good out of politeness. Watch live how they navigate. Note every hesitation, every missed click, every drop-off. Those observations are worth gold, not the technical perfection of your prototype.
Put your web application online
Deploy your web app
Bolt offers one-click integrated deployment via Netlify. Your application is available online on a subdomain, with the option to connect your own domain. The process takes a few minutes from the interface.
For tracking user behavior, connect PostHog or Mixpanel from day one. This data will be essential to understand what works and to know what to keep or rethink before you move to a native version.
Bolt application vs native mobile application: the red line
It has to be said clearly, once more. Your Bolt application is a web application. It opens in a browser. It does not install. It does not appear on the App Store or the Google Play Store. It does not benefit from organic search on the stores, what is called ASO (App Store Optimization), which for many applications represents between 30% and 60% of their downloads.
A native mobile application in React Native is something else. It installs. It sends native push notifications. It uses the phone's full features. It is listed on the stores, with all the acquisition levers that implies.
If your market validates your concept and your growth depends on the stores, the Bolt web app is a validation step. It is not a final product. The right sequence: validate with Bolt, move to native with a team that knows mobile specifics.
Export your code and plan what comes next
Metrics to watch from the start
Four indicators to track from the first days of your launch:
- Activation rate: how many signed-up users complete your app's main action?
- D1 and D7 retention: how many come back the next day, the following week?
- Paid conversion rate, if you have a paywall
- The most frequent exit points in your journey
This data builds your case for what comes next, with an agency, an investor, or your own conviction.
Getting onto the stores with Expo: what Bolt really allows
This is a point most tutorials miss. Bolt offers an official Expo integration, which means you can scaffold a React Native application directly from the Bolt interface, not only a web app.
The limit is elsewhere: Bolt runs in WebContainers in your browser, which does not allow compiling iOS or Android binaries. To publish on the App Store or Google Play, you have to go through EAS (Expo Application Services), an external service that handles compilation, certificates, and signing. It is not one click. It requires configuration, an Apple Developer account at $99/year, and a few hours of setup.
With motivation, it is doable. The result will not be as refined as a native application built from scratch, but it is a real path to land on the stores from a Bolt prototype.
When to call in an expert
Two signals show that your prototype has done its job and that it is time to move to the next step.
First signal: your users engage and ask you for a real application. They want notifications, they want to download something, they look for your app on the stores. The market is telling you the product holds.
Second signal: your web acquisition costs are heavy, and you need the stores' organic acquisition to be profitable. At that stage, a native React Native application with a solid ASO strategy can divide your CPA by two or three.
What to know: today, many of our clients arrive at firstapp with a Bolt prototype under their arm. They have validated their idea, they have their first users, and they want us to take over from there. That is exactly what we do. We take what Bolt generated, we assess what is solid, and we build the native React Native version on that base. No need to start from zero: your validation work has value, and we build on it.
Bolt as a springboard, not as a destination
Bolt is a powerful tool for what it does: quickly generate a working web application with a real backend, from a natural-language prompt. For a founder who wants to validate a concept without writing a line of code, it is hard to find better in terms of control and flexibility.
Its limits are just as real. An application created with Bolt is not a native mobile application. It does not benefit from ASO. It cannot rival the user experience of a well-built React Native application. The Expo path exists for the most motivated, but it has its own constraints.
What we regularly see at firstapp: founders who built their prototype on Bolt, validated their market, and come to us so we take over from there. It is a starting point we value. We pick the project up where it is, we build the native version, and we handle everything that makes a mobile application actually take off: onboarding, paywall, ASO, retention.
If you are in that situation, let's talk.
Frequently asked questions about Bolt
Can you create a real mobile application with Bolt.new?
Bolt offers an official Expo integration that lets you scaffold a React Native application from its interface. But compiling iOS and Android binaries does not happen inside Bolt. It requires EAS (Expo Application Services), an external service that handles certificates and signing. You also need an Apple Developer account ($99/year) to publish on the App Store. So it is technically possible with motivation, but it does not happen in one click. For a native application built properly, with a real onboarding, paywall, and ASO strategy, professional React Native development remains the most solid path.
What is the difference between Bolt and Lovable for creating an application?
Both tools create web applications from natural-language prompts. Bolt offers more technical flexibility, runs a complete development environment in the browser, and allows a full export of the code, an advantage if you plan to hand the project to developers. Lovable produces more polished interfaces visually and integrates more simply with Supabase. For a non-technical profile looking for a fast, clean result, Lovable is often a better fit. For a semi-technical profile who wants more control, Bolt is a serious alternative.
Is Bolt free for creating an application?
Bolt offers a free plan that includes a limited volume of tokens. Those tokens are consumed on every generation, change, and fix. The free plan is enough to explore the tool and test a first simple prototype. To build a complete MVP with several features, paid plans are necessary. They start around $20/month. A precise, complete initial prompt consumes far fewer tokens than a series of successive corrections.
Do you need development skills to use Bolt?
Bolt does not require you to know how to code to generate an application. However, the tool is aimed at a slightly more technical profile than competitors such as Lovable. Understanding basic notions such as the difference between frontend and backend, what an API is, or what it means to deploy an application makes the experience considerably easier. For pure non-technical people, Bolt is usable, but the learning curve is steeper.
Can you export the code generated by Bolt?
Yes, and it is one of Bolt's distinct advantages compared with other tools. Bolt allows a full export of your application's source code, front and back. You can take that code and deploy it in any environment, hand it to a developer to evolve it, or migrate it to a stack better suited to your production needs.
How do you monetize an application created with Bolt?
Bolt lets you integrate Stripe to handle one-time payments and subscriptions. The technical setup is doable through prompts. However, the monetization strategy - price, plan structure, the moment the paywall appears in the journey - stays entirely on you. Those are the decisions that determine whether your application converts at 1% or at 8%. The technical integration is the simple part. The strategy is what makes the difference.