There are moments when an idea feels urgent: a simple fix, a product that could matter, a small dream of launching something real. This guide meets that feeling with a clear path. It explains how an AI-driven builder turns plain language into a working product in minutes.
The reader will see what to expect, how fast progress appears, and why this approach speeds development from concept to a live preview. The guide outlines the flow: describe the app, let the system generate code and structure, then iterate with focused prompts.
Whether someone has never written code or prefers direct control, this tool supports both paths—choose frameworks, connect source control, or edit code. A short video shows the core steps in under two minutes, making it easy to follow along and test an idea quickly.
Outcome: faster prototypes, simpler hosting via the platform, and a measurable route from plan to published app. We position this guide to build confidence and momentum for ambitious creators.
Key Takeaways
- This guide shows what you will build and how quickly you’ll see progress.
- Describe the idea, let the builder generate a structure, then refine with prompts.
- Works for non-technical founders and engineers who want speed plus code control.
- Built-in cloud features remove separate hosting or database setup for early previews.
- A short demo video helps translate ideas into a testable app in one session.
Introduction: What You’ll Build Today and Why Bolt.new Is Ideal for Beginners
This introduction sets one clear promise: in about 10–15 minutes you will create and preview a simple, working app that proves an idea.
Follow a few focused steps: open the site, submit a clear prompt, sign in with Google, GitHub, or email (no credit card), and preview the result. The flow keeps the initial lift low while maximizing early results.
Why this guide matters: it removes local setup and complex tools so newcomers can learn by doing. Each step you take shapes the system’s output; being specific in your prompt speeds progress and improves results.
The underlying model interprets prompts dynamically, so individual builds may vary. That variability is useful: small differences spark new ideas and rapid iteration.
- See a live interface fast and test core interactions.
- Gain confidence with a repeatable, low-friction workflow.
- Use the quickstart guide to follow exact steps and accelerate your learning: quickstart guide.
What Is Bolt.new and How It Works Right Now
An AI-first builder turns short descriptions into a working prototype in minutes. Describe your goal and the platform composes architecture, UI, and runnable code using language prompts and model-driven agents.
AI-powered building for websites, web apps, and mobile apps
This system supports marketing sites, transactional web apps, and mobile experiences via Expo. In practice, a single prompt can generate a landing page, a simple CRM, or an Expo-based mobile preview.
The engine uses large language model agents to interpret intent and map it to implementation details. That mapping includes routes, components, and basic data wiring so you can preview a functional app fast.
Who it’s for: non-coders and experienced developers alike
Non-technical builders get a low-friction path from idea to prototype through chat-driven prompts and previews. Experienced developers can select languages and frameworks, connect GitHub, and edit code directly for deeper customization.
- Quick orientation: a short overview video shows the workflow in about two minutes.
- Scope matters: pick a clear, small feature set to reach a usable preview faster.
- Practical payoff: this approach compresses development cycles so teams focus on product value, not initial plumbing.
What You Need Before You Start
A quick setup check—browser, sign-in, and a brief time block—makes the first build painless. This section lists the practical prerequisites so ambitious creators spend energy on product choices, not configuration.
Browser, time estimate, and a quick note on sign-in
Browser: Use a modern Chromium-based browser—Chrome, Edge, Brave, Opera, or Vivaldi—to ensure the editor, preview, and chat all work on one page.
Time: Reserve about 10–15 minutes to submit your first prompt and see a working app appear. This short block avoids context switching and accelerates learning.
- Sign-in: Sign in or sign up with Google, GitHub, or email—no credit card required—so your workspace ties to your identity.
- The first step is simple: open bolt.new, draft a clear prompt, and submit; the platform handles environment setup.
- Expect on-screen usage indicators so you can monitor LLM usage and manage resource consumption.
With these basics—browser, small time commitment, and lightweight sign-in—users can focus on the product. The preview page loads quickly, letting teams test ideas and refine the next step without extra tooling.
Kickstart Your First App with a Single Prompt
Kick off the build with a single, precise instruction that describes the app’s essential workflow. This short approach turns intent into a working preview quickly. Keep your request focused: list main screens, auth, and any saved items you need.
Open the site and submit a clear prompt
Open the workspace, paste a concise prompt such as “Build a pet name picker app with sign-up and saved favorites,” and click submit. Sign in if requested and complete a short survey. Wait about five minutes while the system plans and generates the first pass.
Preview, interact, and measure results
When the preview loads, click, enter text, and navigate to test core flows. This hands-on check shows early gaps and immediate value. Use the selector tool to point to elements you want to change.
Iterate fast with simple chat prompts
Request small updates in plain language: “change the title,” “add pagination,” or “improve the empty state.” For deeper work, open code view to edit directly. Keep prompts incremental to reduce rework.
| Step | Action | Typical time |
|---|---|---|
| 1 | Paste focused prompt and submit | 1–2 minutes |
| 2 | Sign in and complete survey | 1 minute |
| 3 | Wait for generation and test preview | ~5 minutes |
| 4 | Iterate via chat or code edits | Varies by changes |
Tokens Explained: Plan, Measure, and Optimize
Tokens are the hidden meter that measures the model’s work as it reads, reasons, and builds.
How tokens power read, think, and build phases
Tokens reflect the computational effort behind each action. Longer prompts or larger contexts consume more tokens and increase cost.
Read, think, build: a single request may require the system to read your prompt, plan the change, and then construct code or UI. Each step uses tokens differently.
Turn on visibility and check your balance
Toggle “Display token usage in chat” under General settings to see usage per prompt. Visit Subscription in your profile to review remaining balance and plan limits.
Right-size prompts to conserve tokens
- Prefer short, targeted prompts to reduce token burn while keeping steady development.
- Batch related edits to avoid repeated context building and redundant reads.
- Choose agents intentionally: higher-quality agents use more tokens; lean agents are faster for layout experiments.
“Track tokens early — visibility changes behavior and yields smarter experiments.”
Practical takeaway: monitor token usage, plan work in chunks, and adjust prompts as you learn. Over time, teams find a cadence that balances speed, precision, and cost while advancing the app.
Data, Users, and Authentication in Bolt Cloud
Project data and account access are provisioned automatically, so you can focus on product logic rather than setup.
The platform creates a managed database for application data (favorites, records) and a separate store for authentication when your build requires them. Authentication and data stores are ready to inspect—no separate setup needed.
Explore records in a table view to verify entries and relationships. Use built-in logs, edge functions, and centralized secrets to troubleshoot events and protect sensitive values.

Inspecting and managing users
Built-in user management lets teams review accounts, trigger password resets, and monitor activity during early tests. Claim the database if you prefer to connect to Supabase for external tooling and long-term visibility.
“An integrated data layer reduces overhead while keeping an exit ramp to established services.”
| Feature | Purpose | When to use |
|---|---|---|
| database | Store app records and relations | Immediate prototypes and saved items |
| authentication | Manage users and sessions | Sign-up, login, access control |
| table view | Inspect and edit records | Troubleshoot and verify flows |
| Integrations | Claim DB, connect to Supabase or apis | External tooling and production migration |
- Tip: Batch schema changes and inspect the table after tests to avoid surprises.
- Tip: Use edge functions and secrets to handle sensitive logic off the client.
Editing Your App: Prompts, Code View, and Version History
Editing an app blends conversational prompts, direct file edits, and safe rollbacks into a single, predictable workflow. This gives teams clear control and keeps iteration fast.
Use the selector and chat for precise UI copy
Point with the selector to an on-screen element—change “Pet Name Picker” to “PetPick”—then submit a chat request. This tool finds the right location without guessing file names.
Open code view for exact edits and token savings
Switch to code view when you know what to change. Edits in App.tsx do not consume tokens: search for text, update the string, save, and preview instantly.
Restore a previous build with Version History
Open the project title menu to view past snapshots. Preview a prior state and click “Restore this version” to roll back if recent changes degrade the experience.
Practical workflow: prompt when you want, code when you want
- Prompt for broad or multi-file changes; batch requests to reduce rework and token use. See guidance on prompting effectively.
- Code for surgical fixes and precise strings; search the repo to speed edits.
- Version restores act as safety nets—use them freely during experiments.
“Keep change discipline: describe outcomes clearly to reduce ambiguity and wasted iterations.”
Bolt Agents and Maximizing Token Efficiency
Picking an agent is a strategic decision that balances prototype speed against production quality.
Claude Agent suits larger, production-quality work: it produces more complete results and fewer errors but often uses more tokens and takes longer.
v1 Agent is lighter and faster. Use it for layout checks and quick visual trials where token efficiency matters most.
Balancing speed, quality, and token usage
- Select Claude Agent for complex logic and robust UI states to reduce downstream fixes.
- Choose v1 Agent to validate visuals quickly and conserve tokens during early exploration.
- Consider a two-stage flow: prototype with v1, finalize with Claude for stability.
Strategies to reduce errors and rework
Write prompts that include acceptance criteria—describe what “done” looks like in behavior and UI. When errors appear, report expected vs actual behavior in chat; concrete context speeds fixes.
| Agent | Best for | Token use | Typical outcome |
|---|---|---|---|
| Claude Agent | Production features and complex flows | High | Complete, fewer errors |
| v1 Agent | Rapid layout and prototyping | Low | Fast visual checks, more iterations |
| Claude 4.5 variant | Custom balance of speed vs capability | Variable | Tunable trade-offs for development |
“Align agent choice with your goals—speed for exploration, thoroughness for stabilization.”
Publishing Made Easy: Hosting, Domains, and Live URLs
Publishing moves the prototype from private work to a shareable, testable URL in seconds.
Built-in hosting delivers a live page instantly. Deploy directly from the editor and get a free .bolt.host URL for quick demos and stakeholder reviews. This removes friction so teams gather feedback without configuration delays.
One-click publishing to a free .bolt.host subdomain
From the editor you can publish a page with one action. The platform handles SSL, CDNs, and delivery so the first project becomes visible immediately.
Connecting custom domains for Pro users
Pro plans can attach a custom domain for brand consistency and SEO benefits. This makes a website feel professional while keeping management centralized.
Managing your app’s web presence inside Bolt
Domain records, DNS guidance, and deployment history are visible inside the workspace. Teams avoid juggling third-party services and keep operational tasks in one place.
“Deploy fast. Gather feedback. Treat each release as a testable milestone.”
| Action | Benefit | When to use |
|---|---|---|
| One-click deploy | Live .bolt.host page in seconds | Stakeholder demos and quick tests |
| Custom domain | Branding and SEO uplift | Public launches and marketing pages |
| Domain management | Centralized DNS and SSL | Ongoing maintenance across projects |
- Deploy from editor: share a page link and collect feedback quickly.
- No external services: hosting and operational details are managed for you.
- Scale across projects: keep multiple environments for parallel exploration.
Integrations and Services: From Design to Payments
Integrations turn a prototype into a product by connecting design, source control, mobile targets, and payments.
Design hand-off. Designers can push Figma artifacts into the builder so UI components map directly to code. This reduces guesswork and speeds hand-off.
Version control and collaboration with GitHub
Connect source control to enable version control workflows. Teams use branches, pull requests, and code review while keeping edits safe.
Mobile with Expo and payments with Stripe
Export to Expo for device previews and shared code across platforms. Add Stripe to accept payments and test monetization flows inside the same UI.
Extending functionality with APIs
External apis connect analytics, notifications, or search without reworking the core. Use simple api keys and webhooks to extend behavior quickly.
“Integrations let teams move fast without sacrificing governance.”
| Integration | Purpose | When to use |
|---|---|---|
| Figma | Design-to-code hand-off | Early UI alignment |
| GitHub | Version control and backups | Team collaboration |
| Expo | Mobile previews | Device testing |
| Stripe | Payments | Monetization |
- Practical note: combine these services to keep speed and governance in balance.
- Result: faster delivery, cleaner development workflows, and an app ready for scale.
Bolt.new Tutorials
Short, practical lessons guide builders through each action so results appear predictably.
Follow step-by-step builds, videos, and help center guidance to learn by doing. A hands-on example — the Pet Name Picker — walks you from the first prompt to a live URL. It covers turning on token display, testing sign-up, saving favorites, using version history, and publishing.
Watch a concise video guide to see the exact steps and UI interactions. Then use planning exercises to define scope and acceptance criteria before you build app features.
Plan your app, prompt effectively, and use Discussion mode
Practice prompts that spell out UI states and data flows; clear prompts reduce rework. Use Discussion mode to clarify intent before larger edits.
- Explore guided examples from first prompt to publish to internalize best practices.
- Turn on token visibility to interpret consumption and shape iteration cycles.
- Combine small coding edits in code view with conversational prompts for precise results.
Need a quick walk-through? See the video guide for a compact example that maps every step to a reproducible outcome.
“Plan clearly, use examples, and iterate in small steps to move faster with confidence.”
Conclusion
Commit to small, measurable steps: frame a clear idea, run a short build, and refine the result until the feature set fits your goals.
This workflow compresses development time: submit focused prompts, preview a live app, and apply targeted code changes while version history and managed databases keep control intact.
Use integrations—Figma, GitHub, Expo, Stripe—and pick the right agent to balance speed and stability. Track tokens and test data flows so changes expose fewer errors as projects grow.
Start small, publish often, and learn fast. For a step-by-step example, see the build an app guide: build an app guide.
FAQ
What will I build in the "Beginner’s Guide to Building Your First App on Bolt.new"?
The guide walks a beginner through creating a simple web or mobile app—from a single prompt to a live preview. It covers prompt crafting, previewing generated UI, adding basic data and authentication, and publishing to a hosted URL so learners ship a working app in one session.
Why is Bolt.new a good choice for beginners?
Bolt.new pairs AI-driven generation with a low-friction editor, making it approachable for non-coders while retaining tools for developers. The platform automates routine setup—database, auth, hosting—so beginners focus on ideas and UX rather than infrastructure.
How does Bolt.new create apps using AI?
The service interprets a clear natural-language prompt, then generates UI, routes, and basic data models. Users preview and interact with the app immediately, then refine via chat prompts or direct edits. The flow blends AI suggestions with manual control for predictable results.
Who benefits most from this platform?
Both aspiring non-coders and experienced developers benefit: entrepreneurs and product managers iterate quickly without full engineering resources, while developers use generated scaffolding as a starting point and switch to code view when needed.
What do I need before starting?
A modern browser and 30–90 minutes depending on scope. Sign-in is required to save projects and enable cloud features like auth and persistent data. Optional: a Figma file for design hand-off or a GitHub account for version control.
How do I kickstart an app with a single prompt?
Open the editor, enter a clear, specific prompt describing the app’s purpose, pages, and key features, then submit. The platform generates a live preview you can click through and test. Iterate by sending follow-up prompts to refine layout, copy, or behavior.
What should I expect from generated previews?
Previews are interactive prototypes: clickable UI, simulated navigation, and sample data. They show a working baseline but may need iterations to meet precise requirements. Individual output varies with prompt clarity and chosen agent.
How can I refine features after generation?
Use short chat prompts to request changes—swap colors, add fields, change text—or open the code view for direct edits. The platform supports a hybrid workflow: use prompts for speed and code edits for fine-grained control.
What are tokens and how are they measured?
Tokens measure compute used by AI for reading the prompt, reasoning, and building the app. The platform tracks consumption per chat and build. Monitoring token usage helps estimate costs and optimize prompt length.
How do I track and reduce token consumption?
Enable token display in chat to see per-request usage. Keep prompts concise, reuse existing project context instead of repeating details, and prefer focused follow-up messages. Switching agents or simplifying requested changes also lowers usage.
How does data and authentication work in the cloud environment?
Each project can provision an automatic database and built-in auth. The editor creates data tables for app models and a user table for authentication, making it simple to store user-generated content and manage sign-in flows without separate setup.
Can I inspect and manage app data?
Yes—open the table view to explore records, modify schemas, and manage users. For advanced needs, you can optionally connect the project database to Supabase or another external service.
How do I edit app content and structure?
Use the selector tool and chat prompts to change UI copy and properties. For code-level changes, open Code View to edit files directly. Version History lets you restore prior builds if an edit introduces issues.
What is the recommended workflow between prompts and code edits?
Use the AI chat for rapid iteration and prototyping; switch to code view when you need deterministic behavior, custom integrations, or to avoid token costs. Commit to version history regularly to track changes.
What are agents and how do I choose between them?
Agents are underlying AI models (for example Claude or a platform v1 agent) that trade off speed, quality, and token efficiency. Choose a higher-quality agent for complex logic and a cost-efficient agent for small tweaks and rapid prototyping.
How can I minimize errors and rework when using agents?
Be explicit in prompts, provide examples or schemas, validate outputs in the preview, and iterate in small steps. For critical flows, switch to code view and write deterministic tests or manual checks.
How simple is publishing and hosting?
Publishing is one click: apps deploy to a free .bolt.host subdomain for instant sharing. Pro plans allow connecting custom domains and managing SSL. The hosting panel centralizes live URLs and deployment history.
Can I connect a custom domain or use a professional host?
Yes—Pro users can attach custom domains and configure DNS. For external hosting or advanced deployment workflows, export code and integrate with platforms like Vercel or Netlify via GitHub.
What integrations are supported for design and development?
The platform supports Figma import for design hand-off, GitHub for version control and collaboration, Expo for mobile builds, and Stripe for payments. These integrations help bridge design, code, and production services.
Are there step-by-step learning resources available?
The help center offers written guides, video walkthroughs, and step-by-step builds. Follow guided examples to learn prompt patterns, use Discussion mode for clarifying choices, and consult reference projects to jumpstart ideas.
How should I plan my app and prompt effectively?
Start with the core user story, list essential screens and data fields, and write a concise prompt describing features and sample content. Prioritize minimum viable functionality, then iterate on visuals and integrations.


