You Built an App With AI and Don't Code — How to Put It Online

If an AI tool built you something that runs on your machine and you don't write code yourself, the shortest path online is: zip your project folder, drop it into the Runny AI Agent, and answer a question or two. The agent reads your files, works out what language and dependencies you need, writes the container configuration for you, builds it, and watches the logs afterwards to check it actually stayed running. No Dockerfile, no terminal, no server. There are real limits — it handles Python, Node.js, and prebuilt .NET, and the zip has to be 50 MB or smaller — but for most AI-generated projects that's enough.

What you actually do#

The page walks you through it in three steps, and the first one is the one worth getting right. In its own words: "Start with the same folder you run locally. Keep the essentials and drop bulky, auto-generated files."

Specifically, it asks you to "include source code, config files, and anything the app reads at runtime" and to "skip build outputs and caches like bin/obj, dist, node_modules, venv, or .venv."

That instruction is doing real work. Those excluded folders are usually the majority of your project's size — node_modules alone can be hundreds of megabytes — and they're all rebuilt during deployment anyway. Deleting them before zipping is often the difference between fitting in 50 MB and not.

So: copy your project folder, delete those directories from the copy, zip it, drop it in.

What the agent does with it#

It's not matching your project against a list of templates. It reads the actual files — your imports, your config, your entry point — and writes a deployment plan from what it finds. That's why it copes with projects that don't follow a standard layout, which AI-generated ones frequently don't.

Two things are worth knowing about how it behaves:

It will ask you for secrets rather than guessing. If your app needs an API key or a bot token, the agent pauses and asks you to supply it. You paste the value into the app's configuration, not into your code — which is the correct place for it, and means the key isn't sitting in your project files where it could later be shared by accident.

It checks its own work. After the container starts, it reads the real logs looking for crashes rather than assuming success. If it finds a problem, it gets one attempt to diagnose and fix it.

You get a limited number of builds per day — the panel shows how many you have left — so it's worth zipping carefully rather than uploading five times in a row.

Three things that surprise people#

These aren't your fault or the AI's. They're facts about running in a container that don't apply on a laptop.

1. Files your app writes will disappear. If your app saves data to a file next to its own code — a small database, uploaded images, a JSON file it updates — that data is discarded every time the app is updated. It doesn't error; it just starts empty again.

The fix is a volume, which is storage that survives. If your app stores anything users care about, this is the one thing to sort out early. More on where data goes.

2. It has to listen in the right place. Web apps need to accept connections from outside their container, and code written for a laptop often listens only to itself. Nothing checks this for you, so the deploy succeeds and the page never loads. If your app deploys "successfully" but the link doesn't work, this is the most likely reason — and it's a one-line change your AI tool can make if you tell it the app is running in a container.

3. "Running" doesn't mean "working". The status tells you the container is up, not that your app is functioning. Open the link and use the app. That's the only check that proves anything — more on verifying properly.

When you'll need to learn one technical thing#

Honest limits, so you don't lose an evening:

• Other languages. The zip path handles Python, Node.js, and prebuilt .NET. Anything else needs a Dockerfile — a short configuration file your AI tool can write if you ask for one.
• Bigger projects. Over 50 MB zipped, after removing the caches, means a different route.
• Something that won't build. If the build fails twice for the same reason, ask your AI tool to look at the error rather than re-uploading.

None of these require becoming a developer. They mostly require asking your AI assistant a more specific question — and telling it where the app will run, which is the single most useful thing you can add to a prompt.

What it costs#

On a new account, the first app you deploy gets a short trial run — genuinely enough to see whether your thing works. It then stops until you add credit, and from there it runs continuously, billed by the minute for the CPU, memory, and disk you allocate; the number of apps isn't itself a billing unit. Current rates are on the pricing page.

The practical advice: start at the smallest size. It's easy to increase later, and most small apps need far less than people assume.

FAQ#

Can I deploy an app if I don't know how to code?#

Yes, for common cases. Zip your project folder and upload it — the agent works out how to build and run it, and asks you for any keys it needs.

What should I include in the zip?#

Your source code, config files, and anything the app reads while running. Leave out node_modules, venv, dist, and bin/obj — they're rebuilt for you and take most of the space.

Which languages does it support?#

Python, Node.js, and prebuilt .NET. Other languages work but need a Dockerfile.

Where do I put my API key?#

The agent asks for it during deployment and stores it as app configuration. Don't put it in your code files.

Why did my app's data disappear?#

It was saved inside the container rather than on a volume. Add a volume and point your app at it.

My app says it's running but the link doesn't work.#

Usually the app is listening only to itself instead of accepting outside connections. Ask your AI tool to bind to all interfaces.

Wrapping up#

The gap between "it works on my laptop" and "it's on the internet" is smaller than it used to be, and for a small AI-built project it's mostly one careful zip. Strip the caches, upload, hand over the keys when asked, then open the link and actually use the app rather than trusting a green status. The two things worth learning early are where your data lives and how your app listens — everything else you can pick up when you need it.