Your App Can Now Go Live: Deployment for Every Stack, In Your Own Cloud

Until recently, VrittOS took an idea all the way to reviewed, tested code on GitHub pull requests — and stopped there. The last step, actually putting the app on the internet, was a guide and a wish. That gap is now closed for every stack the platform generates: websites, full-stack apps with databases, and — as of this week — Flutter mobile apps.

Where your app can live

• Vercel or Netlify — websites and front-end apps. Free tiers cover most projects.
• Railway — full-stack apps with a backend. VrittOS provisions the project, a private Postgres database, your app service and its public URL, all inside your Railway account. Usage-based, typically $5–20/month.
• Render — container hosting with a free tier. Your GitHub Actions builds the app into an image using a Dockerfile we've proven in real containers, and Render runs it at a public URL.
• Appetize — for Flutter apps, something better than a build log: a link where anyone runs your Android app in a browser-streamed emulator. No install, nothing to sideload — send the link to a tester or an investor and they're using your app seconds later.

The rules that don't bend

Deployment touches your hosting accounts and your money, so the design is deliberately conservative:

1. Your accounts, your bills. Everything is provisioned in your Vercel/Railway/Render/Appetize account. Hosting is billed by the provider directly to you; VrittOS never charges for running your app, and the panel shows honest monthly cost ranges before you pick.
2. We never store your credentials. The token you paste is used once — to set things up and to write the deploy secrets into your own GitHub repository — then discarded. There is no token column in our database, by design.
3. A pull request is the switch. Connecting a host opens a PR that adds the deploy workflow to your repo. Nothing deploys until you merge it. After that, every change that reaches your production branch goes live automatically — still through the same review gate every other change passes.
4. "Live" means we checked. After each deploy, VrittOS fetches your app's URL and only reports Live when it actually answers. The one-time go-live charge (10 credits) happens on the first verified live deploy — never for a deploy that didn't work. Re-deploys are free.

When a deploy goes wrong

Real deploys fail in real ways — a missing runtime secret, a server binding the wrong port, a build tool mismatch. Two things ship alongside the integrations:

• A Fix button on the failure itself. When a deploy fails verification, the panel shows what went wrong and a Fix it for me button: the AI investigates the failure, fixes the root cause in your code, and opens a pull request — merging it redeploys automatically. And if the site is up but wrong, you can describe the problem in a sentence and get the same treatment.
• Secrets handled before they hurt. Generated apps refuse to boot without their security keys — correctly. So hosting setup now pre-configures those keys with securely generated values in your hosting account, and the setup PR lists any variable you still need to set yourself, before the first deploy instead of after a crash.

The reliability work you don't see

The same weeks that shipped these integrations also hardened the machinery under them: a watchdog that detects and restarts stalled AI runs within minutes (and turns a second stall into an honest failure with options, not silence), automatic recovery of pull requests that lost their link to a story, Dart code generation wired into every Flutter build, and a free get help from our team hatch on stories that have outlived the automation. When something does slip through, our policy is simple: we fix the cause for everyone, and the affected founder gets credits back with an apology — not a status page euphemism.

What this means if you're a founder

The distance from "I described my product in a paragraph" to "here's the URL, try it" is now one unbroken, reviewable path: brief → design → stories → code on PRs → tests → live app on your own hosting. The parts you own are the approvals; the parts the AI owns are verifiable at every step.

See your idea live: start a free trial, build something, and press Publish. If you've already got a project at code-complete, the Publish panel is waiting — it knows your stack and recommends where to host it. The full journey is described in the AI-native SDLC.