Your AI builds it.
One prompt ships it.
You already build apps by asking an AI assistant. Connect Roar AI and the same assistant puts them online, with a database and a link to share, without leaving the chat.
From the chat
to a live app
Nothing to install and no settings to learn. You keep working the way you already do, and say when it should go live.
Connect once
Add Roar AI to your assistant and sign in. There is no key to copy or keep safe.
Build as you do today
Keep asking your assistant to write and change the app. It lives on GitHub, like any project.
Ask it to go live
One prompt deploys it with a database and an address. Ask again to change it, fix it or roll it back.
Writing the app
was the easy part
An assistant can write an app in an afternoon. Running it needs servers, a database, an address and somebody keeping an eye on it. That is what Roar AI is, and now your assistant can drive it.
How App Hosting worksA home for the app
It runs on our servers in the region you pick, with scaling and a certificate handled.
A Postgres database
Created beside the app and connected to it, with backups twice a day.
An address people can open
An address on roarai.app straight away, or your own domain when you are ready.
A new version on every push
Change the code, push it to GitHub, and the new version goes live.
Give your Claude
200+ models
to work with
Not every job needs your best model. Labelling thousands of rows, summarising a folder of notes, pulling fields out of invoices: your assistant can hand the work to a fast, cheap model through Roar AI and bring back the results, on the same account and the same bill.
Running a job needs an assistant that can run code, such as Claude Code, Cursor or Codex.
Picks the right model
Your assistant sees every model on your account, with what it can do and what it costs, and picks a fast, cheap one for bulk work.
A key with a cap
It makes a key that can only call models, with a spending cap, so a long job cannot run away with your budget.
One API for all of them
Every model answers on the same OpenAI-compatible API, so trying another one is a one-line change.
Run your apps
in plain language
Deploying is the start. The same assistant can fix, change, watch and pay for what it runs, through over 80 tools.
Deploy the intranet repo from GitHub with a database, and tell me when it is live.
Why did the last deploy fail? Show me the build log and fix it.
The new version is broken. Roll it back to the last one that worked.
Set FEATURE_SEARCH=on on the intranet app and redeploy it.
How many bookings came in this week, by clinic?
Put the intranet on intranet.ourcompany.com.
Run the cleanup script every night at 2am.
Is anything erroring on the support bot? Show me the last hour of logs.
What did we spend on AI this month, and which app spent the most?
Your assistant,
your permissions
The MCP server calls the same API the console does. Nothing your assistant does goes around the rules you already have.
Only what you can do
Every call goes through the same permission and ownership checks as the console. An assistant can never do more than the person behind it.
Asks before it deletes
Every tool says whether it only reads, changes something, or cannot be undone, so your assistant knows when to check with you first.
People keep the big calls
Inviting colleagues, topping up credit, editing rows in your database and restoring a backup stay in the console, with a person.
Disconnect any time
Remove Roar AI from your assistant and it lets go of its sign-in straight away. Anything already granted expires within the hour.
Connected in
about a minute
Pick your assistant. It signs you in through the browser the first time, and remembers it after that.
Claude
Add Roar AI from Claude’s connector directory. Claude asks you to sign in the first time.
Connect nowClaude Code
For the terminal. Run this once:
claude mcp add --transport http roar https://cloud.roar-ai.com/mcpCodex
Registers Roar AI and signs you in. The CLI, IDE and desktop app share it.
codex mcp add roar --url https://cloud.roar-ai.com/mcp && codex mcp login roarQuestions about the MCP
What is MCP?
The Model Context Protocol is the open standard AI assistants use to work with other services. Our MCP server is what lets your assistant deploy apps, read logs and check spend on your Roar AI account.
Do I need to be a developer?
No. If you can describe what you want to your assistant, you can ask it to put it online. The app’s code needs to be on GitHub, and a coding assistant like Claude Code or Cursor can put it there for you.
Which assistants work with it?
Claude, ChatGPT, Cursor, VS Code with Copilot, Claude Code, Codex and OpenCode, and anything else that accepts an MCP server address.
Do I need an API key?
No. Most assistants sign you in through your browser the first time and remember it, so there is nothing to copy or keep safe. A key is only for an agent that runs on its own with nobody there.
What can my assistant see?
Only what you can, in the projects you can reach. It cannot see another workspace, and an assistant you connected to read usage never sees a deploy tool at all.
What does it cost?
Connecting costs nothing. What your assistant runs for you, apps, databases and AI calls, is billed at the same prices as when you do it yourself in the console.
Your assistant
is ready to ship
Connect Roar AI and put your next app online from the chat you are already in.