Run Chatify on your own computer.

Run the app and store its files on your own computer. Cloud model providers still receive the requests you send them. Install the small uv tool once, then one command installs and starts Chatify.

Open a command window

Mac: press Cmd+Space, type Terminal, press Return. Windows: open Start, type PowerShell, open it. Linux: open Terminal from your app menu.

Install uv once

uv is the small installer Chatify uses. You only do this the first time.

Mac / Linux:

curl -LsSf https://astral.sh/uv/install.sh | sh

Windows PowerShell:

powershell -ExecutionPolicy ByPass -c "irm https://astral.sh/uv/install.ps1 | iex"

When it finishes, close and reopen the command window, then check it worked:

uv --version

You should see a version number. If you see "command not found", the window was not reopened yet.

Run the one command

Paste this and press Enter:

uvx --from chatify-app chatify start

The first run downloads Chatify; after that it starts in seconds.

Check it worked

Your browser opens your new app automatically. A short guide appears on first launch: you can skip it at any time, and it never locks the app while it is open. Ask a question in the box at the bottom of the Chat page.

If something goes wrong

"uv is not recognized" or "command not found": close and reopen the command window after installing uv, then try again.
No curl, or the installer script fails: install uv another way: pip install uv on any Python setup, or brew install uv on a Mac with Homebrew, then run the one command above.
The browser did not open: look for a line like http://127.0.0.1:8080 in the command window and open it yourself.
Windows shows a red policy error: use the PowerShell command exactly as written above; it includes the permission for this one install.
Still stuck? The GitHub page has the full setup notes: github.com/Harsh23Kashyap/Chatify ↗

Good to know before you start

The easiest setup is Build with AI. On the Configuration page, choose Build with AI and describe the app you want in plain words. It drafts the whole setup for you to review. Nothing is saved until you confirm.
Questions go in the box at the bottom of the Chat page. Type your question there and press Enter. Answers cite the sources they used.
Add a public web page as a source: in Add sources, choose "Add a web page" and paste the address. Chatify stores that page's text and answers over it.
Files can be up to 8 MB each. If a file is too large, remove it and choose a smaller one; large PDFs often work once figures are removed.
Local storage is not the same as offline processing. App files and saved chats are stored on the machine running Chatify. If you choose a cloud model, requests can include your question, conversation context and source passages. Browser dictation may use an online speech service. Choose a local model and avoid online features when you need offline processing.

Try Build with AI here first ↗ · Source and docs on GitHub ↗