Bolt.new
StackBlitz's browser-based AI app builder that runs a full Node.js environment via WebContainers and ships a working full-stack app from one prompt.
Who it's for
Non-developers and solo founders who want a working full-stack app on a live URL straight from a prompt, without installing anything locally.
Standout details
- WebContainers runs a complete Node.js environment inside the browser tab itself, so there is no local setup, just a prompt and a URL
- Free plan gives 1M tokens a month capped at 300K a day, enough to see whether the workflow suits you before paying
- Pro at $25/mo raises the allowance to 10M tokens and drops the daily cap; Teams at $30/user/mo gives each member their own 10M tokens rather than a shared pool
- One-click integrations cover Supabase, Netlify, Stripe and GitHub, and a Figma file can be dropped straight into the chat to build from
- The same token meter bills for the mistake and for the fix, so a session that goes wrong costs twice
What it is
Bolt.new is StackBlitz’s chat-to-app builder. You describe what you want, it writes a full-stack React and Node app, and runs it live in the browser through WebContainers, the same in-browser Node.js runtime StackBlitz built for its online IDE. There is no local install and no build step to configure, so the app is running before a normal project setup would have finished.
StackBlitz has said Bolt reached $20M in annualized revenue within about two months of launch, and in 2026 announced a partnership bringing Bolt to Microsoft Azure and Microsoft 365 for enterprise distribution, alongside its existing AWS Marketplace listing. That is a distribution deal, not a change of ownership: Bolt still ships under its own name, at its own domain.
Who should use it
Non-developers and marketers who need a working prototype, landing page or internal tool today, and solo founders validating an idea before they write a line of code themselves. Lovable sits in the same lane with a similar credit-metered model; the practical difference is Bolt runs the whole stack inside the browser tab itself, where Lovable deploys to its own hosting from the start.
Developers can use it too, for the same reason: scaffolding a CRUD app or an auth flow by prompt is faster than typing it by hand. It stops being the right tool the moment the remaining work is the kind of business logic or edge-case debugging that AI generation handles poorly.
What stands out
Everything runs through WebContainers, which is why there is no setup step at all. Your browser tab is running Node itself, not streaming from a remote sandbox, so even the free plan is a real full-stack environment rather than a stripped-down demo.
The tier ladder is a token ladder. Free gives 1M tokens a month, throttled to 300K a day so nobody burns the monthly pool in one afternoon. Pro at $25/mo removes that daily throttle and multiplies the pool to 10M. Teams at $30 a seat is not just Pro with collaboration bolted on: each seat carries its own 10M tokens rather than drawing from one shared pool, so the allowance scales with headcount instead of being split between however many people are on the plan.
The integrations are the other reason it is fast: Supabase for a database, Stripe for payments, Netlify for a second deploy target beyond Bolt’s own hosting, GitHub for a repo you actually own, and a Figma drop-in so a design file becomes the build reference instead of a text description of one.
The verdict
For the first pass at an idea, Bolt is hard to beat on speed: prompt in, app running, no setup in between. That is real, and it is why founders reach for it before anything else.
Go in expecting the token economics to bite. Superdesign’s 2026 review of user reports describes one developer spending over 500,000 tokens fixing a single authentication bug, with debugging routinely costing more tokens than building the original feature. Product Hunt reviewers, in the same round of 2026 coverage, put their own failure-repair share of spend at close to half. That is how you end up with what reviewers now call the “70% wall”: Bolt gets you most of the way to a working app fast, and the last stretch of real business logic is where the token bill and your patience both run out.
If you need something that looks finished for a demo or an MVP pitch, that ceiling does not matter. If the app has to be correct with real users and real money attached, budget for a developer to finish what Bolt starts, rather than expecting the tool to close that last 30% on its own.
Honest verdict
Bolt gets a full-stack app onto a live URL faster than almost anything else in this category, because WebContainers removes the environment setup entirely. The catch reviewers keep returning to is the token meter: it charges the same for Bolt's own mistakes as for your prompts, and multiple 2026 reviews describe the bulk of a session's spend going to repairing errors Bolt introduced. Buy it for a landing page, a prototype, or an MVP demo. Do not buy it expecting a finished production app, because once the work turns into real business logic or handles real money, plan on a developer to take it the rest of the way.
Published 30 Aug 2026