Apps of the future, launched today

Créez votre propre app, par vous-même.

Apportez l’idée. Orbit crée le produit, le lance sur votre domaine et continue de l’améliorer avec vous.You build the product, put it on your own domain, and grow it from there.

Your first launch is free

Build it with the Claude or ChatGPT subscription you already pay for. I add nothing on top, and the first app stays live for free.

Your own domain

Bring your own and point it at the free app. Adding a domain is not an upgrade — HTTPS and sign-in come wired.

Your own data

It sits in a database nothing else shares. Not a few rows inside somebody else's table.

The parts you build in

One app. Five surfaces you can see and change.

Open one, change it, and the rest keeps up.

Scratchpad

Start with the idea

Notes, links, the pitch you typed at 1am. Build straight off it.

01 / 05
Scratchpad surface
Data model

Draw the data

Draw the tables. They run as Postgres, and your rows survive a deploy.

02 / 05
Data model surface
Workflows

Wire the logic

Say what happens when somebody books. It runs on a server, not a tab.

03 / 05
Workflow surface
Mockup

Draw the screens

Draw each screen, wired to your data. What you draw is what ships.

04 / 05
Mockup surface
Launch

Put it live

One deploy. A URL, HTTPS and a database, on a domain you own.

05 / 05
Launch surface

The Orbit flywheel

An app is never done.

Version one is a guess. You find out what you built the week somebody uses it. What you learn on Friday, you change on Monday.

Explore
Ideas, notes and links, in one place.
Shape
Decide what you are building, and what you are not.
Build
The screens, the data, the logic.
Launch
Your own domain, with hosting and HTTPS.
Grow
Watch people use it. That tells you the next thing.

Grow feeds Explore, and round again. The parts a prototype skips are the parts that make the second turn cheap.

Five ways in, one app

Build from wherever you already work.

Claude, Codex, WhatsApp, the terminal, the browser. All of them read and write the same app. No export, and no moment where somebody says to open the real tool now.

Claude and the Orbit studio editing the same app side by side

Claude Code · the same app open in the studio

MCP

Codex building an Orbit app through MCP

Codex · Orbit MCP

MCP

A WhatsApp chat that builds an Orbit app

WhatsApp · screens come back as a picture

chat

$ orbit login

$ orbit apps create --title "Bright Meadow Dental"

$ orbit deploy create --app bright-meadow-dental

live at https://brightmeadow.dental

orbit · the CLI

terminal

The Orbit studio in the browser

launch.orbitapps.io · the browser studio

browser

Claude, or any MCP agent

Point it at Orbit and it works on the real app, not a sketch of one.

Codex

Same connection, same app. The backend comes with it.

WhatsApp

Describe it in a message. Screens come back as a picture. Tap once and they are live.

The CLI

Create, edit and launch without leaving the terminal.

The browser studio

The whole builder in a tab. Nothing to install, and the preview follows you.

Read this before you sign up

Where Orbit is worse.

Orbit is one bet about how an app gets built: data first, composed, not written. Every bet costs something. This one costs you four things. Better to know now than in week three.

01

Orbit does not write code

You compose UI elements, integrations and a database that already exist. There is no escape hatch to a file. Cursor, Claude Code and ChatGPT will write you anything. Orbit will not.

02

No free AI credits

Lovable, Bolt, Replit and Emergent bundle model credits into the price. Orbit does not. Bring your own key and it costs nothing. Otherwise every run comes off your balance.

03

It only runs on Orbit

Your data model, your data and your domain are yours. The app runs on Orbit's hosting, and nowhere else. There is no export button. Ask, and I will build one.

04

No genie in the bottle

One prompt gets you a screen. It will not get you a million-dollar app. The rest is work, and it is yours. The most Orbit can do is make it a joy.

What the bet pays

The parts you would have built next.

A page Google can find. Screens that update while you watch. A real app on a phone. A way for other software, and other people's AI, to use it. Each one is usually its own project. Here they all come from the app you already described.

postgres

There is a real app underneath

Its own database, work that runs on a server while nobody is watching, and a way for people to sign in. The parts a demo skips.

llms.txt

Found by Google, and by ChatGPT

Every page is built on the server, so a search engine reads the real thing. Your app ships the files that crawlers and answer engines look for.

websocket

Nobody has to hit refresh

When something changes, every open screen shows it straight away. Orbit still checks who is allowed to see it before it sends anything.

android · pwa

The same app, on a phone

Your app becomes a native Android build, with the icon taken from your own mark. On any phone, an iPhone included, people install it from a link.

/openapi.json

Other software can plug in

Your data gets a proper API, and the document that describes it, both written for you. Anything you marked private stays out of both.

POST /mcp

Your customers' AI can use it

Your app becomes a set of tools an AI assistant can call. Each person connects with their own key, so the assistant sees only what they can see.

Pricing

Your first launch is free.

After that there is no plan. Everything comes out of one balance, and nothing on it expires.

What you pay for

Your first live appfree

Its own Postgres, sign-in, and a domain you own. It stays free for as long as you keep it up, and it is not a trial.

Each deployment after thatby the hour

You pay for the hours it is up, and stopping it stops the meter. A live site and its preview count as two.

Models, search and imagesas they run

Drawn from the same balance, with no keys to keep and no second bill. Bring your own key and it draws nothing.

Credits never expire, they are refundable, and nothing auto-charges.

See what each one costs

Straight answers

Six things people ask first.

Your move

Build the one you keep talking about.

Your customers will use it. So will their agents. Building an app is hard enough. It doesn't have to be dull. The first one is free, on a domain of your own, and nobody is going to ask you for a card.

Créé avec Orbit