Viewfy Mini. Free and open source.
He drafts the reply.
Nothing leaves your browser
Replies and posts on
Mini reads your X account to learn how you sound, understands the page you are on, and drafts the message in your voice. The model is the one already inside Chrome, so there is no account, no server and no API key.
Apache 2.0. Read every line before you run it.
He sits in the composer you already use. One click and the message is written in your words, on your own machine.
Three steps. You still post.
Load it in Chrome.
Open chrome://extensions, turn on Developer mode, hit Load unpacked and pick the folder. The first time you open the popup, Chrome fetches the model. It is a multi gigabyte download, once for the whole browser, and every extension that uses it shares the same copy.
Learn your voice.
Open your own profile on X, scroll a few posts into view, and press Learn from you. He reads that page only, and only when you press the button.
Click him where you type.
The badge sits in the composer. One click and the message is written in your voice, in place. Click again for another take, or type a few words first and he builds from them.
What he never does.
No account.
No sign in, no email, no domain, no card. Nothing to create and nothing to cancel.
No server.
The writing happens in Chrome's own model, on your machine. Viewfy never sees the thread or the draft.
Nothing read in the background.
He reads the thread when you ask for a draft, and your profile when you press Learn from you. He does not watch what you type or where else you browse.
No dashes, no tells.
Drafts never carry an em dash or an en dash. The prompt bans them and the code enforces it.
What it runs on.
The model belongs to Chrome, so these are Chrome's requirements for it, not ours. Mini adds none of its own.
- Browser
- Google Chrome 138 or newer, on desktop. Chrome for Android and iOS cannot run the model. Some Chromium forks ship a stub that only echoes your text back, and Mini tells you when it sees one.
- System
- Windows 10 or 11, macOS 13 Ventura or newer, Linux, or ChromeOS from platform 16389 on a Chromebook Plus.
- Free space
- At least 22 GB on the volume that holds your Chrome profile. If you later drop below 10 GB Chrome removes the model, and fetches it again once there is room.
- Graphics or memory
- More than 4 GB of VRAM, or 16 GB of RAM with four or more CPU cores.
- Network
- An unmetered connection for the first download only. After that he drafts with no connection at all.
Not sure about your machine? Open chrome://on-device-internals and read the device performance class.
Questions.
Do I need an account?
No. There is no sign in, no email and no domain. The only gate is the one time model download, and Chrome does that for you from the popup.
Where does the AI run?
Inside your browser. Mini uses Chrome's built in model, Gemini Nano, through the Prompt API. Your posts and the threads you read never leave the machine, so there is nothing for us to store.
Which sites?
, for now. Replies, the reply modal and the home composer. No DMs.
Does he post for me?
No. The draft lands where you were already typing. You edit it there and press Post yourself. Nothing is ever sent for you.
How is this different from the full Viewfy?
Mini drafts where you already are. The full Viewfy finds the threads where your buyers are asking, while you are not on those sites, and queues the drafts for your approval.
Mini drafts where you already are.
The full Viewfy finds the threads for you, on Reddit, LinkedIn, X, Threads, DEV and Quora, and has the draft waiting in the toolbar when you open Chrome.