A backend that comes assembled.
Every app needs the same plumbing: a database to store things, a way to sign users in, email that lands, a place to take money, a folder for uploaded files, an AI model or three, a domain to live on. Building all of this used to mean buying five products, learning five dashboards, and writing the glue between them.
somewhere.cloud is that plumbing, pre-wired. Your agent writes what makes your product different — the screens, the logic, the brand. We run what every product needs. One login. One bill. One place to ask when something goes wrong.
You describe it. Your agent ships it.
Your agent creates the project
You say what you want in the chat you already use — ChatGPT, Claude, Codex or Cursor. Your agent reads our guide for agents and creates a project for the app.
You approve one sign-in
When it needs your account, it stops and shows you a code and a link. You open the link in your own browser, check the code matches, and click Allow. Nothing signs you in behind your back, and you do this once.
It builds, runs and checks it
Your agent writes the app, runs it, and looks at the result in a real browser — the same way you would — fixing what it finds before you see it.
Your app is live
It puts the app on its own live link and sends you the URL. The database, sign-in, files, email and AI are already wired; there is nothing for you to connect.
Once real people use it, every change your agent makes gets a private preview link with its own copy of your data, so nothing it tries can touch what your users see. When you are happy, it ships the exact preview you reviewed — nothing else.
Build your product. Not the plumbing.
Every app needs the same plumbing, and every team rebuilds it: sign-in and password resets, data only its owner can read, payments that never double-charge, jobs that retry safely, uploads that stay private. Here it is already built, so your agent starts on your product.
Nothing to stitch together. Nothing to keep in sync.
We handle the plumbing. You build the product.
Any of them. Same result.
In ChatGPT or Claude with a workspace of their own, your agent keeps working between conversations. It signs in once — with your approval — and stays signed in, so the next thing you ask for picks up where the last one left off.
In Claude Code, Codex or Cursor on your own computer, the same sign-in opens the approval page in your browser. Once you have clicked Allow, your agent can create, build, check and ship from that folder.
Using Claude or ChatGPT in the browser without any of that? Add Somewhere as a connector once, in the app's settings, and build from the conversation. Every path reaches the same database, files, sign-in, email and deploys.
How to add the connector →See it built end-to-end.
emailsomewhere is a complete email product running entirely on somewhere.cloud — the same capabilities your agent gets, no special access, nothing custom underneath. If you want a worked example of what a real app here looks like before you ask for yours, this is it.
Compare current database behavior, limits, portability, and attached PostgreSQL options in the somewhere.cloud vs Postgres guide.
- Inbox and outboxReceives mail at its own domain and sends from it — the platform's email, in both directions.
- Per-user dataA dozen database tables, each fenced to the signed-in user by the platform, not by hand.
- AI insideSorts and drafts with the built-in models; queues the outbound batches in the background.
- Live updatesThe inbox refreshes as mail arrives, over the same realtime channel any app here can use.
Three places to look.
Every capability, every detail, at /docs. Written for coding agents, but you can read it too — or just ask your agent, which already has.
Floating button in the bottom right of every dashboard page. It knows the reference and which page you are on. Ask anything.
For "this is broken" or "could you build X" — ask your agent to file a ticket, or use Feedback in the dashboard. Real humans read these and reply by email.