Getting started
From an empty SD card to a running device
Flash Raspberry Pi OS, install the Freymi agent from your browser, pick a project. About 20 minutes — most of it waiting for the card to write.
What you need
Nothing exotic. If you already have a Pi on the shelf, you are most of the way there.
Raspberry Pi
Pi 4 or Pi 5, a power supply and a microSD card of 4 GB or more. A display if your project draws something on screen. Compare models →
Network
Wi-Fi or Ethernet. The device only needs outbound HTTPS: no port forwarding, no static IP, no VPN.
A browser
A free Raspberry Pi account and Raspberry Pi Connect. Nothing to install on your own computer, no SSH client, no terminal.
Step 1 — Flash Raspberry Pi OS
Insert the microSD card and open Raspberry Pi Imager — free for Windows, macOS and Linux.
- Device: your Pi model. OS: Raspberry Pi OS (64-bit) with Desktop — the graphical session is what lets you reach the Pi's screen from a browser later.
- Before writing, open the settings (gear icon) and fill in hostname, username + password and Wi-Fi.
- Recent versions of Imager also have a Raspberry Pi Connect tab in those settings. If you see it, turn it on and paste an auth key from your Raspberry Pi account — the device then links itself on first boot and you can skip Step 2 entirely.
- Write the card, put it in the Pi, power it up. The first boot takes a minute or two.
Step-by-step with screenshots: Getting started — Raspberry Pi documentation.
Step 2 — Turn on Raspberry Pi Connect
Open Connect, sign in with your Raspberry Pi account and name the device. It then appears at connect.raspberrypi.com, reachable from anywhere — including from behind a router you do not control, which is the usual situation in a guesthouse. Unplug the monitor and keyboard; you are done with them.
Full reference: Raspberry Pi Connect documentation.
Step 3 — Register the device in Freymi
In the Freymi dashboard open Devices → Add device, name it and copy the device token.
The token is the device's identity: it is what the agent uses to report status and to learn which project it should run. Keep it private — the 8-character project code is public, the device token is not.
Step 4 — Install and start the agent
Open connect.raspberrypi.com, click Connect → Screen sharing next to your device, and you are looking at the Pi's desktop in your browser. Open a terminal there and run:
curl -fsSL "https://freymi.app/api/public/system-files/freymi-agent/freymi.py" -o freymi.py
FREYMI_DEVICE=<device-token> python3 freymi.pyThat is the whole installation, and it is the only time you will see a terminal. The agent is a single Python file with no dependencies — no pip, no virtualenv, nothing to keep up to date. On start it compares its own SHA-256 checksum against the published one and replaces itself if a newer version exists, then asks the API which projects belong to your account.
Step 5 — Write your first program
The hardware is done. Go back to the dashboard, create a project and describe what the device should do — in your own words, the way you would explain it to a colleague. For example:
“I want a screen with the weather forecast for Prague — large colourful icons, today plus the next three days.”
Freymi turns that into a working project: it writes the files, builds a release and makes it available to your devices. Be concrete about the things you actually care about — the city, the layout, how often it should refresh — and the first result will land much closer to what you had in mind.
Select the project for your device and the agent picks it up on its next check, downloads the release and starts it: main.py for Python projects, index.html served locally for kiosk screens. Output streams back to the project's Logs tab and the device reports started, running, stopped or error on the Devices page.
Want something changed? Adjust the prompt, publish a new version, and the device updates itself — you never touch the terminal again.
The endpoints behind all of this are public and documented: Freymi API documentation.
What it costs
Credits pay for generating projects — turning your description into working code. Everything else about the setup on this page is free.
Ten credits a week, free
Every registered account gets ten credits a week, at no cost. That is enough for roughly two smaller projects — describing one, generating it and getting it onto a device. You do not need a card to start, and there is no trial that quietly ends.
Top up when you need more
If a week runs short, buy credits and spend them when you like. Pay as you go: no monthly plan, no tier to choose, no minimum. In a month where you build nothing, you pay nothing.
The device stays yours
We don't sell you hardware and we don't rent it to you. The Pi is yours, the SD card is yours, and the agent that runs on it costs nothing. If you walk away, there is nothing to return and nothing to cancel.
Official Raspberry Pi resources
Everything in Steps 1 and 2 is standard Raspberry Pi setup. If you get stuck, these are the sources we follow ourselves.
Getting started with Raspberry Pi
The official walkthrough of imaging a card and the first boot, with screenshots for every Imager setting.
Raspberry Pi Connect
How Connect works, what screen sharing needs, and how to manage your devices and auth keys.
Raspberry Pi software
Download Imager for Windows, macOS or Linux, and browse the official operating system images.
Common questions
The things people ask before their first device is up.
Have a Pi on the shelf?
Sign up, write a prompt, and see it running on the device the same afternoon.