Initial setup and your first model
Install o42, connect an AI model (a Gemini key or a local model), and find where the rest of your settings live.
o42 is a sovereign desktop app: it runs and stores all your information locally, and you choose which AI model powers it. Nothing starts until you give it that brain — this first tutorial gets you ready to work.
What you need before you start
- A clean o42 install, with no model connected yet.
- A Google AI Studio key if you’re going with Gemini, or willingness to download a local model — you don’t need both.
Connect a model
The first time you open «o», or when you start your onboarding journey, o42 shows you a window to connect a model. You have two paths.
Option A · Gemini key
Paste a Gemini key into the field and confirm. It’s the fastest way to get started. If you don’t have one yet, the link in the window takes you to Google AI Studio to generate one in seconds.
Option B · Local model
For maximum sovereignty, choose “Choose a local model”. o42 downloads and runs the model directly on your machine — it downloads once and stays available offline, without sending your data to any provider. Ideal if you work with sensitive information.
With a model connected, «o» can think. Notice the model selector: it appears in chat, in goal creation, and in several other flows — you decide, at every moment, which AI to use.
Where your settings live
All your configuration lives in Settings, accessible from your avatar button in the top right.
- Private info — manage your Gemini, Anthropic, and OpenAI keys. You can hold several and switch providers whenever you want: o42 is model-agnostic.
- Local model and command-line tools — the local model card and the Gemini CLI, Claude Code, and Codex integrations live further down the same screen.
- Public info — set your display name and avatar: that’s how «o» recognizes you, and how you appear if you join an organization.
Next: Meet «o»
With your model connected, the next step is meeting the agent that accompanies you across the app.