FAQ

You need a machine that runs Docker and ten minutes of patience. The installer is one command. The AI key goes into a settings screen if you can install an app, you can run Frely.

Any VPS with 2GB RAM handles a solo freelancer’s workload comfortably. A home server, NAS or Raspberry Pi 4 works too. If it runs Docker, it runs Frely.

Anything OpenAI-compatible: OpenRouter (one key, dozens of models), DeepSeek, Azure OpenAI, or a local LLM behind an OpenAI-compatible endpoint. You choose the model per cost and quality.

Data is stored locally in your Postgres database. Only text you explicitly send for AI processing goes out over your own API key, directly to your chosen provider. There are no Frely servers in the middle. There are none.

Yes a Postgres dump is a complete, portable export of your entire business data. No proprietary formats, no export fees, no “please contact support”.

As much as your own server is. Client work never leaves your infrastructure unless you explicitly send text to the AI provider you picked.

Zero for everyday drafting: the built-in local model runs on your machine. If you plug in an API key for heavy project work, you pay your provider directly, typically a few dollars a month at solo volume. Frely itself is free during beta and stays free for core features.

The beta is small on purpose: each new user gets a real setup conversation with the builder, not a docs link. Invites go out in waves as the install experience gets smoother.

Nothing happens to it it was always yours, on your server. Take a dump, keep running your current version, or migrate wherever you want. No lock-in exists to escape from.