How it works Frely

From zero to your first AI-drafted proposal in one evening. No DevOps degree required if you can run Docker, you are 90% done.

0typical install time

0command to install

0RAM is enough

0monthly AI cost, local model

Step 1 · five minutes

Install with one command

Clone the repo, run docker compose up -d. Frely, its Postgres database and the voice service come up together on any VPS, home server or NAS.

git clone https://github.com/frely/frely.git
cd frely && docker compose up -d

Step 2 · two minutes

Go local. No key needed

Frely ships with a lightweight local model: everyday drafting works offline, right on your machine, no account and no bill. Want an extra brain for heavy project work later? Plug in any OpenAI-compatible key in Settings → AI: OpenRouter, DeepSeek, Azure or a local LLM endpoint.

Step 3 · ten minutes

Feed it your history

Paste three to five of your best proposals and one or two real client threads. Frely learns your structure, your tone, your wins. From that moment drafts sound like you on a good day.

Every day after

Work from one window

New job post? Generate a proposal. Client agreed on scope in chat? It becomes a task with a date. Friday evening? Open Analytics and see exactly where the week went and what to bill.

What you need

Any machine with Docker. A $5 VPS, a home server, a NAS, even a Raspberry Pi 4 handles it.
Nothing else for everyday work. The local model handles drafting offline. Optional: an OpenAI-compatible key (OpenRouter gives you dozens of models with one key).
Ten minutes of patience. The installer does the rest. If you ever add an API key, it goes into a settings screen, not a config file.

Ready to stop writing proposals by hand?

Join the waitlist and get your invite when the next wave opens.

Join the waitlist