Quick start

You're about to turn an idea in your head into a real, running app. No servers to set up, no terminal, no config files — and no long form to fill in before anything happens. You describe the idea, you answer a few questions, and the app starts building while you're still talking.

Before you start

Here's all you need, and exactly what it costs.

• An account. Sign up at pikkustudio.app. That's the only setup step.
• Free credit to start. Signing up puts free credit in your account, and you pay as you go after that, in credits. Following this guide uses a little of it.
• No installs. You can do everything in the browser. There's an optional command-line tool for developers who prefer it — see the CLI reference — but you don't need it here.

When you first sign in, you land in the Console — your home base for building, deploying, and watching your app.

Step 1: Start a project

New project gives you four ways in. Take the first one — Describe it — and write your idea the way you'd explain it to a friend:

An app where I log my day — what I got done, how I felt — and can share a day with my therapist.

You can also start from a template, import a GitHub repository that already holds a Pikku app, or work from the CLI. Any of those land you in the same place.

While your workspace boots, you'll be shown a handful of things this app could grow into and asked which ones you like. It takes ten seconds and nothing gets built from it — it's remembered for later, so when your own thinking heads somewhere you already liked, the system knows.

Step 2: Have the conversation

This is the part that's different from other AI builders, so it's worth knowing what to expect.

Pikku Studio Cloud doesn't hand you a plan to approve. It asks one question at a time, and before each question it tells you what it's already decided:

Sharing with a therapist and a partner are different things — one is professional, one is personal, and they shouldn't see the same amount.

I'm making sharing per-person and per-thing, not all-or-nothing, and off by default.

Who writes the entries — just you, or other people too?

When the answers can be named, they arrive as buttons rather than a blank box. Your job is to correct it where it's wrong, which is much easier than specifying it from scratch. Nothing is locked in — every decision is reversible with a sentence.

Annotated screenshot of the Chat: an assistant turn stating a consequence, stating a decision already made, and a question card with two labelled option buttons. Callout: "It decides, you correct — one question at a time."

Step 3: It starts building

There's no Build button. Once enough is settled for a first milestone to be real, it's dispatched and you get one line saying so — "I've started building the daily log while we keep going."

The conversation doesn't stop. You'll be answering questions about the next piece while the first one lands.

Two places to watch it:

• The Weave — every function, route and screen appearing as it's woven. During the first build this is the main event.
• App preview — the running app itself, the moment there's something to click.

You may be asked first whether you want to see a couple of looks for a screen before it's built. Say yes and real options get sketched for you to pick from.

Step 4: Read what you agreed to

Open Knowledge and you'll find every decision written down — what your app is about, who uses it, and why each call was made.

This isn't a transcript. It's the specification the builder is actually working from, in plain language. If something in there is wrong, say so in chat and both the note and the app follow.

Step 5: Publish it

Everything so far happened in a private sandbox with no route to the internet. Publish is the one step that changes that: it checks your work, commits it, and ships it to a stage.

Send it to staging first, click around, then promote to production. Out of the box production stops and asks for your approval before it goes live, even after every check has passed — and you can roll back any deployment in one click. If you'd rather it went out on its own, you can turn auto-deploy on for that stage.

That's the whole loop. Talk, it decides, it builds, you publish.

Where to go next

• Core concepts — the handful of words you'll see everywhere.
• Chat — how the conversation works, and the other two agents behind it.
• FAQ — straight answers on cost, lock-in, and whether it's safe to ship.