Make room for your idea
Sign up with your phone, name your project and pick a technology and database. Everything is prepared in seconds, with nobody to wait for.
brewInTH.You bring the idea. AI brings the code.
We give it a place to grow.
From an AI conversation to a live application. Copy your project link to your AI and let it upload, build, wire the database and launch. Hosting, databases and deployment, all in one place.
Let AI help you create. Let brewInTH take care of your project’s new home, from servers and databases to a console you actually understand.
Sign up with your phone, name your project and pick a technology and database. Everything is prepared in seconds, with nobody to wait for.
Copy the connection link and token to your AI. It uploads the code, builds it, wires the database and reports back in your language.
Your site goes live on its own URL. Browse release history, roll back, watch storage and bandwidth from one page, and upgrade when you are ready.
Anyone who builds websites, apps or APIs with AI and wants them live without managing servers, Docker, DNS or databases themselves. That includes non-programmers and developers who value their time.
No. There are two ways. With an agent-style AI such as Claude Code, Codex or Cursor: create a project on the web, copy the link and token and paste them to the AI; it reads the project guide, uploads the code, builds it, wires the database and reports the URL back. With a plain chat AI such as ChatGPT, Claude or Gemini: download the code as a ZIP, or copy it, and hand it over on the “Upload and go live” page; the platform works out the technology, creates the database and puts the site online.
Backups are an add-on on any package: automatic daily database backups (keeps 7), manual backup/restore/download, and /bw-data volume backups. Without it the platform keeps no copies, so export your own.
Any AI that writes code. Agent-style tools that can call HTTP — Claude Code, Codex CLI, Cursor, Windsurf and other tools that support REST or MCP — deploy for you from start to finish. For a plain chat that cannot call APIs (ChatGPT, Claude or Gemini in the chat window) there is the “Upload and go live” page: hand over the ZIP or paste the code the AI wrote, and the platform puts it online.
One project, 20 MB of files, one 512 MB database, 1 GB of bandwidth per month, a brewInTH address with a name you choose, and HTTPS issued for you. An app on the free package stays up; it sleeps only after three days without a single visitor and wakes when someone opens it. Enough to try your first idea without a credit card.
Each project runs in its own sandbox with its own database and its own CPU, memory and storage limits. An AI receives only the permissions you grant for that project, and you can revoke them at any time.
Mostly not. The platform passes the port in PORT (8080 by default), provides the database connection under the names code usually reads (DATABASE_URL, DB_HOST and so on) and offers persistent storage for files that must stay, such as uploads or an SQLite file. Two things to know: files an app writes outside persistent storage are gone at the next release, and code should not hard-code its database at localhost. The guide your AI reads covers all of this.
Yes. Every project gets its own HTTPS hostname with REST, WebSocket, server-sent events and gRPC (HTTP/2 end to end), a PostgreSQL, MySQL, MariaDB or MongoDB database, Redis for caching and queues, scheduled jobs and outbound access to any public port except 25 (send mail via 465/587 or a provider API). A backend without a home page sets a health check path such as /health in the advanced publish options, or lets the AI pass healthPath.
brewInTH serves from machines in Thailand; each project runs on one machine as one instance with the CPU/RAM of its package (no multi-region, no autoscaling yet). Free: requests up to 20 MB, responses awaited for 2 minutes, 32 WebSocket connections, apps sleep without visitors, and serving stops at the monthly bandwidth quota. Paid packages: 400 MB requests, 10-minute responses, hundreds of WebSocket connections, and serving continues up to 150% of the quota with email warnings at 80% and 100%. Each project can have its own Redis for caching and queues; there is no built-in object storage yet (external services work).