You asked an AI assistant to build something, and it did. There is a working app on your machine: maybe a website, a small tool for your team, an online store. Then comes the question every builder hits next:
"How do I put this online so other people can use it?"
If you have never done it before, the honest answer is: traditionally, with a surprising amount of homework. This post walks through what that homework is, and then through the newer option — letting the same assistant that built your app publish it too.
What "putting it online" traditionally means
An app on your laptop only works while your laptop is on and only for you. Putting it online means getting it running on a computer that is always on, reachable at a web address, secured with HTTPS so browsers trust it.
Doing that yourself usually looks like this:
- Pick a hosting company and make an account.
- Install their command-line tool.
- Write a configuration file describing your app.
- Tell the platform how to build and start it.
- Move your passwords and keys over safely.
- Point a web address at the result.
- Push, wait, and — often — watch it fail.
- Read a log, work out why, change something, try again.
None of these steps is impossible. But every one of them assumes vocabulary and patience that have nothing to do with the thing you actually built. If you made your app by describing it in plain language, being handed a build log at the finish line is a strange ending.
The newer option: ask the assistant
AI coding assistants like Claude Code and Codex can do more than write code. Through an open standard called MCP (Model Context Protocol), they can connect to outside services and operate them on your behalf — including hosting.
With an agent-native host, the deployment conversation looks like the building conversation:
"Put this online, please."
The assistant packages your project, sends it to the platform, and waits while it is built and started. If something goes wrong along the way, the assistant is told what and can fix it and retry — without you needing to read anything. When it is done, you get back a working HTTPS address you can open and share.
The important shift is who absorbs the failures. Deploys fail all the time, even for professionals; the difference is whether the failure lands on you or on software that knows what to do with it.
What to check before trusting any host with your app
Whether you deploy by hand or through an assistant, some things are worth verifying about wherever your app will live:
- Is the address checked before it goes live? A deploy that "succeeds" but serves an error page to visitors is worse than one that fails loudly. Prefer platforms that verify the URL answers before switching a new version on.
- Can you go back? Updates go wrong. Every previous version should be kept so a bad update can be rolled back to the last one that worked.
- Do your secrets stay secret? Passwords and keys should never end up in a log — or in your chat transcript.
- What is decided up front? Some choices — like whether your app has a database, or built-in sign-in — may need to be made when a project is created. Know which decisions are one-way doors.
- What does it cost to find out? Getting a first app online should be free. You should not need a credit card to learn whether the workflow suits you.
That checklist is, not coincidentally, a description of how Rightside works: releases only activate after the address is verified, every version is kept, secrets are redacted before they reach a log or the assistant, the database and sign-in choices are made clearly at the start, and getting an app online is free.
Honest limits
If you are evaluating this workflow, you should also know what it does not do yet. Rightside is currently a private beta — access is granted by invitation, not public sign-up. Your app gets its own generated HTTPS address; pointing a custom domain of your own at it is a next milestone, not something available today. We would rather tell you that here than have you discover it after deploying.
The short version
- Putting an app online traditionally involves a dozen chores that have nothing to do with what you built.
- AI coding assistants can now do those chores for you, end to end, through MCP.
- Judge any host — agent-operated or not — on verification before going live, the ability to roll back, and how it treats your secrets.
If you built something and want it online, request access to the beta — then ask your assistant, in plain language, to put it online.