A reader wrote in this week describing the Terminal as “a place that I fear to tread”. He got there anyway, and then suggested the manual should explain this properly for people who are new to it.
He was right, so I have done two things: written it down, and removed most of what there was to write down. MojoPad 1.59.0 sets this up for you. You will not need to type a command anywhere.
Why there is a second app at all
A model that writes is a large piece of software in its own right — several gigabytes of it. Putting one inside MojoPad would make every download heavier for everybody, including the many people who never touch these features, and it would be a worse model than the one you can choose for yourself. So MojoPad borrows one from Ollama, a free app whose whole job is to keep models on your Mac and let other apps use them.
Ollama is the only thing you install. MojoPad will not download or run an installer for you, and that is deliberate: the promise this app makes is that nothing happens to your Mac that you did not ask for, and an app that quietly fetches and runs programs has given that promise away. That one step is yours. Everything either side of it is handled in the app.
Your first model, in about two minutes
- Ask MojoPad to do something with AI.
⇧⌘Afor Ask Your Wiki, or select a few words and choose ✦Explain. If there is nothing on this Mac to answer with, MojoPad says so and offers to set it up. You can also go looking for it: Set Up Local AI… in the command palette (⇧⌘P). - Click Get Ollama. Your browser opens at the download page. Download it, drag Ollama to Applications like any Mac app, and open it once.
- Go back to MojoPad. It has already noticed. Leave the setup window open while you install — MojoPad watches for Ollama appearing and moves on by itself. Nothing to click to say you are done.
- Click Download the model. MojoPad looks at how much memory your Mac has and picks a model that suits it, so there is nothing to choose and no way to pick one that is too big. It tells you the size first, and the progress bar is a real one.
- You should see “Ready”. That is it, permanently.
Things worth knowing
You do not need a second model. If you have read that this needs an “embedding” model — that is a real distinction, but not your problem. MojoPad carries its own small one for finding pages by meaning and offers it separately when it is first needed. The download above is the writing half, and it is the only one you have to think about. (Some older messages in the app said otherwise and sent people to a command line for nothing. Those are fixed.)
“Ollama is not running” is different from “Ollama is not installed.” MojoPad now knows which, and says so — it used to give the same advice for both, which meant telling people to install something already sitting in their Applications folder. If it is installed but asleep, there is a button to start it.
The download can be interrupted. Stopping costs you nothing; starting again picks up where it left off rather than beginning from nothing.
Answers very slow? The model is probably too large for the Mac to hold comfortably. Run Set Up Local AI… again for a smaller one, or see Bring your own model in the manual, which lets a model elsewhere answer instead.
Want none of it? Settings ▸ AI ▸ Use no AI at all switches off every one of these features and stops MojoPad offering. There is a per-wiki version in the Info palette for a document you would rather no model ever touched.
Where to read more
The manual now has Getting Started with Ollama as its own chapter — the one above, in full — and Local AI remains the deep chapter: what the two kinds of model actually do, choosing between them, what stays on your Mac versus what can be sent elsewhere if you allow it, and how to switch any of it off.
Thanks to the reader who asked. If you get stuck on any step here, reply and say where — the step that trips people is the one worth fixing next.
— Mark