Your first conversation

Follow these steps

1. 1

   Install the app. Ginu is in a closed beta. The Download page takes an email and offers a place; invited Windows testers receive an installer link by email, and the Mac app follows the same way. Run the installer and open Ginu like any other app.

2. 2

   Accept the agreement. The first launch opens with the terms and the privacy notice. Sending basic usage reports is off until you turn it on. Tick the box and press Accept and continue; Decline and quit leaves nothing behind but a record that you were asked.

3. 3

   Let the prerequisites check run. Ginu checks what it needs on this computer and checks again by itself, so something you install while the screen is open shows up without a restart. git is required, because memory, projects and Work are stored with it. tmux is optional: without it only Shell is off, and it does not apply on Windows. If git is missing, Windows offers an Install button; elsewhere the screen shows a command you can copy.

4. 4

   Have one AI CLI ready, and sign in to it. Ginu talks through Claude Code, Codex or the Gemini CLI and has no model access of its own, so it uses the account you already have. The screen shows a card for each one: Install where Ginu can do that for you, Sign in when it is installed but signed out. Sign-in opens that CLI’s own login, and Ginu never sees the credential. Continue unlocks once git is present and one CLI is ready.

5. 5

   Choose which CLI to talk through. The next screen lists the CLIs it found and the ones it can drive. Pick one; you can change it later in Settings → Engine.

6. 6

   Import, or skip it. Ginu can scan for instructions and memory another CLI already holds, so it starts out knowing some of your conventions. Nothing is imported unless you select it, and Import nothing is a normal answer.

7. 7

   Finish. Ginu is ready in text mode at this point, and nothing speaks until you press Start conversation. Either ask to be asked a few profile questions by voice later, or skip and fill them in under Settings → Profile.

8. 8

   Let the speech models arrive. The first launch downloads a few gigabytes of them; the prerequisites screen states the figure when it is known. You can type to Ginu meanwhile. They stay on your computer afterwards, so later launches skip the download.

9. 9

   Put earphones in, and split your input from your output. Never a wireless headset as both microphone and speaker: while its microphone is open the whole device drops to call quality, and Ginu holds the microphone open for the whole session so you can interrupt it. Use the built-in or a wired microphone for input and the headset for output, and choose them in Settings.

10. 10

    Allow the microphone, then talk. Your computer asks the first time Ginu opens the microphone; if nothing asks and Ginu stays deaf, check your system’s microphone privacy settings for Ginu. Press Start conversation on Home and say something: “What is in this project?” is a good first question. Then cut it off mid-sentence — it stops, hears you, and answers the new thing.