How it works
From nothing to a working AI command center in one evening. You bring a server and an AI key, Frely brings the rest.
Step 1 · Install
One command on your server
Any machine that runs Docker will do: a five-dollar VPS, a home server, a NAS. The compose file brings up Frely, its Postgres database and the voice service together.
git clone https://github.com/frely/frely.git
cd frely && docker compose up -dStep 2 · Connect AI
Paste your AI key in Settings
Frely speaks the OpenAI-compatible protocol, so almost any provider works: OpenRouter, DeepSeek, Azure, a local model on your own GPU. The key lives in your database on your server. You see exactly what AI usage costs, because you pay the provider directly.
Step 3 · Teach it your style
Feed it a few of your wins
Add two or three proposals that won you jobs, plus one real client thread. That is the whole onboarding. From the next job post on, drafts come out in your voice with your structure, and they get sharper with every outcome you log.
What you need
- A server or VPS with 2 GB RAM and Docker (a $5-10/month VPS is enough to start)
- An API key from any OpenAI-compatible AI provider
- Ten minutes for the first run, no special skills beyond copy-paste
Want help with the first install?
Waitlist members get a guided setup. Join, mention your setup, and we will get you running together.