Zugo.dev's answer

Zugo checks that what it built actually runs before it tells you it is done. Every build boots in a sandbox and reports back: a page that throws on load is marked broken instead of being handed over with a green tick, and a verdict can get worse but never better, so nothing quietly reports clean on a retry.

The second thing is range. Games, marketing sites, apps and multi-page platforms come out of the same prompt box, with real playable game engines rather than a screenshot of one.

The third is that you can leave. The source is readable and editable by hand, every build exports to GitHub, the database is your own Supabase project, and enquiry emails are sent from your own domain on your own quota.

Zugo.dev's answer

Three reasons, and they are all things you can check rather than claims.

You are told the truth about the build. The completion tick waits for the preview frame to answer instead of firing when the model stops typing, so you are not looking at a success badge over a blank page.

You can make more than one kind of thing. Most AI builders do web apps. Zugo does playable games with real engines, marketing sites, apps and multi-page platforms from the same box, so you do not need a second tool for the next idea.

Nothing is locked in. GitHub export, your own Supabase project, your own domain for lead email, your own GA4. If you stop paying, the thing you built is still yours and still runs.

The free tier needs no card, so you can compare before you commit.

Zugo.dev's answer

People for whom the first working version is the expensive part.

Founders and solo makers testing an idea who do not want to spend a month finding out it was wrong. Marketers and agencies who need a landing page or a small campaign tool today, without a sprint. Teachers, streamers and hobbyists making small games. Developers who want a working scaffold in minutes that they can export to GitHub and carry on in their own editor.

They have in common that they can describe what they want in plain words, and that they would rather see it running than read a plan of it.

Zugo.dev's answer

Zugo started in mid 2026 as a small experiment: could you type one sentence and get a game you could actually play, not a screenshot of one.

The experiment worked, and then it taught us the thing the product is now built around. Our first version decided a build had succeeded by testing whether the output contained a closing html tag. A document could match that, be perfectly balanced, and still throw on its first line. It billed as a success and showed a green tick over a blank page.

Fixing that turned into the core of the product: the build is mounted in a sandbox, an error harness is injected ahead of the build's own scripts, and the completion tick waits for the frame to answer. Everything since has followed from the same rule, which is that the product is not allowed to tell you something works when it has not checked.

We write up what we measure, including the parts that do not work yet, on the blog.

Zugo.dev's answer

• React and TypeScript for the editor, built with Vite
• Supabase for Postgres, auth and storage, with row level security
• Vercel for hosting the app, the published customer sites and the serverless functions
• Anthropic Claude models for code generation
• esbuild-wasm to bundle generated React apps in the browser, so a preview does not need a server round trip
• Phaser for the game runtime, self hosted rather than pulled from a CDN
• Stripe for billing
• Resend for transactional and lead email
• Vitest and Playwright for the test suite and headless verification