A personal assistant that lives on your machine, speaks through every messaging app you use, and ships real work — while you sleep, while you cook, while you do anything else.
A three-layer memory — tacit knowledge, daily notes, atomic facts — that persists across months and sessions. No more starting over.
Telegram, Discord, Slack, WhatsApp, Signal, email. One brain across every channel you already check.
Reads files, writes files, runs shells, browses the web, books calendars, sends mail — the same tools you would.
Every complex task becomes a reusable skill. The longer it works for you, the sharper it becomes.
"Every morning at 9, summarize my emails." Plain English, then it just runs — quietly, forever.
Local-first. Your keys, your files, your logs. Swap between 200+ models without a migration.
No signup walls, no SaaS dashboards to configure. One install, one wizard, one agent that's yours.
A single line. Works on Mac, Linux, WSL2 — wherever you already code.
A short wizard. Pick a model. Paste a key. Name your agent whatever you'd like.
From the dashboard, or from any messaging app you connect. It just works.
Replaced my entire Zapier stack in a weekend. Mine reads email, books meetings, opens PRs. Feels like a real teammate.
The messaging gateway is the killer feature. Telegram, Slack, email — same agent, same memory. My cofounders didn't even notice.
Setup took twelve minutes. Now my agent ships code while I sleep. Feels slightly illegal.
No lock-in. No seats. Install and go.
Yes. FreX is local-first — your API keys, your files, your conversation logs all stay on your hardware. The model itself runs through whatever provider you pick (Anthropic, OpenAI, Nous, OpenRouter), but everything around it is yours.
You pay your model provider directly for API usage — typically $5–$30/month for individual use, depending on model. Our tutorial and the software are free. The $15 guide is optional.
No. The tutorial is copy-paste commands. Most people are chatting with their agent inside thirty minutes. You only touch code if you want to write custom skills.
WSL2 is required for Hermes. OpenClaw works natively on Mac, Linux, and Windows via WSL2.
We build in public. The live dashboard shows real revenue, real streams, real traction — updated every few days. Transparency builds trust.
We're weeks from the next milestone. Install FreX, ship real work, watch the counter climb.