Fundaia Case Study, Our Platform to Deploy Apps from GitHub

Project type
Deployment platform
Years
2026
Deliverables
Web dashboard, server and containers, command line, and MCP server for AI agents
Fundaia's dashboard in a browser, with a deployment going through its steps, and the same project on a phone.

Fundaia is our platform for putting applications on the internet. You give it your GitHub repository and it builds it, starts it and leaves it at its own address, without you setting up any server. We built it ourselves, and this website's API runs on it.

A phone on the sofa with the same project in Fundaia.

From code to the internet

Each service comes from a GitHub repository, an image, a template or a file, and its deployment shows which step it is on. If something fails, it says where and why, and the previous version keeps working.

All connected

Ready made databases and tools are added from a catalogue, and the services pass each other their addresses and passwords on their own, with nothing copied by hand. Whatever nobody visits for eight hours goes to sleep and wakes up with the next visit.

A phone on the train with Fundaia's catalogue of templates.
A phone in the kitchen with a service's variables, one of them a reference to the database.
A phone in a coworking space with a database query and its results.
A phone at night with a sleeping service that will wake up with the next visit.

For AI too

Everything done in the dashboard can be done from the terminal, or asked of an AI agent such as Claude Code, which runs it with the same functions as a person.

A laptop in a meeting room with Fundaia's agents page, where an AI agent finds out why a service went down.

How it's built

  • Dashboard

    React and Tailwind CSS, for the dashboard and the public pages.

  • Server

    Bun and TypeScript, with every deployment's progress live.

  • Containers

    Podman without administrator rights, each project on its own private network.

  • Domains

    Caddy and Let's Encrypt certificates, so every service has its own secure address.

  • Terminal and agents

    A command line in Rust and an MCP server for an AI agent to use it.

  • Data

    SQLite for the platform, and databases such as Postgres, MySQL or MongoDB for each project, queried from the dashboard.

In numbers

steps in every deployment, in plain sight
6
tools for an AI agent to run it
290
databases you can query from the dashboard
5
Next case: Boathous