Lovable alternatives, by what you're actually building

Short answer: people leave Lovable for three different reasons, and each points somewhere else. If you want the code in your own stack: v0 for Next.js on Vercel, Bolt for a full-stack prototype running in the browser, Dyad if it must be free, local and open-source. If you want a platform that keeps working after launch: Replit for an IDE with hosting and a backend, WeWeb for a visual frontend on a backend you own, Webflow for a designer-and-CMS marketing site. If the site is marketing and you already work in a coding agent: Sites by Clize, ours, or plain DIY with a static host. The reasoning is below.

The comparison at a glance

First — what you would be leaving

Lovable turns a conversation into a working full-stack app in the browser: it writes the frontend, wires a data layer through Supabase, syncs to GitHub, and hosts the result, all without a terminal. That is a genuinely hard thing to do well, and for going from nothing to something people can click, it is very good. The reasons people look elsewhere are usually specific rather than damning — credit costs as iteration grows, wanting the code to live in an existing repo and review process, needing a piece the platform doesn't cover, or simply not needing an app at all when the job was a marketing site. Stay on it if the thing you're building really is an application, and you'd rather not open a terminal at any point.

v0 by Vercel — UI that fits a stack you already have

v0 generates React and Next.js interfaces with Tailwind and shadcn/ui, iterates on them in a chat, and deploys to Vercel. It is narrower than Lovable on purpose: less "here is your app", more "here is a well-built page or component you can paste into the codebase you already maintain". If your production stack is already Next.js on Vercel, the output lands with almost no translation. Choose it if you have an existing React codebase and want generated UI that belongs in it.

Bolt.new — a full dev environment in a browser tab

Bolt runs the actual toolchain client-side through StackBlitz's WebContainers: it installs packages, runs the dev server, and lets the agent edit a real project in front of you, across a wide range of frameworks rather than one blessed stack. Deployment goes out through its Netlify integration. The flip side of that breadth is that you are steering a real project — it rewards someone who can read the diff. Choose it if you want a full-stack prototype running in minutes and are comfortable driving a real dev environment.

Replit — the IDE that also hosts

Replit is not really a site builder; it is a cloud development environment with an agent in it, plus databases, deployments and custom domains. That makes it the strongest option here when the thing you're building keeps growing after launch and needs a server, a scheduled job, or a database you'll query directly. It also means more surface to learn than a chat box. Choose it if you want one place to write, run, host and keep iterating on something with a real backend.

Dyad — free, local, open-source

Dyad is the option with no platform in the middle: an open-source builder that runs on your own machine, with your own model API keys, leaving the code on your disk from the first prompt. You pay model providers directly and arrange hosting yourself. For people whose objection to the hosted builders is lock-in or per-message pricing rather than capability, that trade is the whole point. Choose it if you want a Lovable-shaped workflow that costs nothing beyond model usage and answers to no vendor.

WeWeb — a visual frontend over a backend you own

WeWeb is a visual builder for the frontend specifically, designed to sit on top of a data layer you control — Supabase, Xano, or any REST API — rather than generating one for you. Teams reach for it when the backend already exists or must stay under their governance, and the part they want to move fast on is the interface. Choose it if the data layer is settled and the UI is what needs building, visually, by people who aren't going to write React.

Webflow AI — the designer's platform, with AI in it

Webflow has been the serious answer for designer-built marketing sites for years: a visual designer with real CSS semantics, a CMS, hosting, and a large template and agency ecosystem. Its AI features assist inside that platform rather than replacing it. If a designer or marketer owns the site after launch and needs to edit content without a developer, that continuity is worth more than generation speed. Choose it if the site is marketing, someone non-technical will maintain it, and you need a CMS behind it.

Sites by Clize — a marketing site shipped from your coding agent

Ours, so weigh it accordingly. Sites by Clize doesn't give you an editor: your coding agent writes the site in your own repo, following a hosted design method — 39 style constitutions, palettes and font pairings, frozen into a DESIGN.md and checked against an anti-slop review — and then Clize ships it. The part that is genuinely different is what "ships" covers: the deploy, the HTTPS certificate, the domain the agent can register for you after quoting the price, and a support@ mailbox on that same host which the agent can read and answer. It is static only — no database, no server-side rendering, no logged-in app — and if you don't want to work in a terminal, one of the browser builders above will suit you better. Choose it if the site is a marketing site, you already live in Claude Code or Codex, and you want it to end on a real domain with a working inbox instead of a preview link. The full walkthrough is here.

DIY — your agent and a static host

The option people forget: have Claude Code or Codex write plain HTML and Tailwind, and push it to Netlify, Cloudflare Pages, or GitHub Pages, whose free tiers are generous and whose reliability is not in question. You get total control and no per-message pricing. The costs are real but unglamorous: design consistency is entirely on you and the model's defaults are bland, and the domain, DNS, certificate and mailbox are four separate errands. Choose it if you have design opinions of your own and don't mind doing the wiring once.

How to choose in three questions

1) Is it an app or a site? An application with logins and data → Lovable, Replit, Bolt, or WeWeb over your own backend. A marketing site → Webflow, Sites by Clize, or DIY. 2) Where do you want to work? A browser editor → Lovable, v0, Bolt, Replit, WeWeb, Webflow. Your own repo and agent → v0's output, Dyad, Sites by Clize, DIY. 3) What has to be true when it's done? If a URL is enough, most of these are fine. If it has to be a domain you own, with a certificate and an address people can write to, check who covers those — usually it is you, in three other dashboards, or it is part of the same run.

$ npm i -g @clize/clize
$ clize login
$ clize claim mystudio --email
$ clize build site start "<your brief>"