What this is
The most direct way to get something you described to an AI onto the open web. You stay inside ChatGPT the whole time it writes the code, a landing page, a calculator, a small single-file app, then Revdoku turns those files into a real website. No IDE to install, no build step, no hosting dashboard. It’s the “I just want a link I can send someone” stack.
When to pick this stack
- You’re not a developer (or don’t want to act like one today) and just need a working thing online.
- The output is mostly self-contained: a static page, a prototype, an interactive demo, a one-pager.
- You want to share a URL today and iterate on it tomorrow without redeploying anything.
How publishing works (Revdoku)
Revdoku takes the files ChatGPT generated and publishes them to a live URL in seconds. You choose how open that link is:
- Public, anyone with the link can open it.
- Password-protected, the link plus a shared password.
- Password + email, visitors enter a password and their email, with optional email confirmation so you can verify the address before they’re let in.
Either way you get analytics on how the page is actually used, and in password + email mode those analytics go per visitor, so you can see which person (by email) opened the prototype and how they engaged with it. Updates republish to the same link, and there’s a free plan to start.
What we’d swap
- ChatGPT → Claude: if the build is code-heavy, or you want to design it live in an Artifact before publishing.
- ChatGPT → Codex: when the job is big enough to want an agent that builds and tests a whole project, not a chat that hands you files.
Common gotchas
- ChatGPT will happily reference external scripts or APIs; for a clean static publish, ask it to inline everything into one self-contained file.
- It’s a prototype path, not a production backend. If you need real auth, a database, or server logic, treat this as the demo and graduate to a fuller stack once it’s validated.