Local AI

MojoPad's AI runs entirely on your Mac through Ollama. No accounts, no API keys, no tokens — and no page content ever leaves your machine.

Setup

1. Install Ollama from ollama.com.
2. Pull a chat model — ollama pull qwen2.5 is a fine default.
3. For semantic features, pull an embedding model too: ollama pull nomic-embed-text (or mxbai-embed-large for higher quality).

MojoPad picks models automatically; override in Settings ▸ AI.

Ask your wiki

⇧⌘A opens the Ask palette. Ask a question in plain language; MojoPad retrieves the most relevant pages (semantically, when an embedding model is present), hands them to the model, and streams back an answer with source chips — click one to open the page it drew from. You can also select text in any page, right-click, and choose Ask Your Wiki about “…”.

The switch at the top of the pane picks what you're asking. My wiki grounds every answer in your own pages, with sources. This page scopes the conversation to the page you have open — including PDF attachments, whose text is extracted on the fly, and it reads the whole document: a 300-page book is indexed into sections on first use (you'll see the progress), and each question retrieves the most relevant passages from anywhere in it. Ask about chapter 14 and it actually reads chapter 14. (Once read, a PDF's text also joins document search for the session. Long-document questions need an embedding model — see Setup.) Anything asks the model directly — its own general knowledge, no retrieval — for when you want a definition, a translation, or a starting point that isn't in your wiki yet. All three stay entirely on your Mac, and MojoPad remembers which mode you used last.

A good answer is worth keeping: press ↳ Save as Page beneath it and the answer becomes a real page — named after your question, with the answer formatted and a footer that records when you asked and links to the source pages. Those are ordinary wiki links, so the saved answer appears in the graph and in each source's Backlinks, exactly like a page you wrote yourself. Answers from Anything mode are stamped as the model's own knowledge instead — so future-you knows which pages came from your notes and which came from a machine.

Summarize and tag

The Ask palette's Summarize button condenses the current page to bullet points — and it works on PDF attachments too. Long documents are summarized in full: the book is read part by part (progress ticks off in the pane), then the part summaries are distilled into one final digest of the whole work. A full-length book takes some minutes of local crunching — start it, make coffee, come back to an executive summary. Suggest tags proposes tags you can add with a click.

Semantic search and Related

With an embedding model installed, MojoPad quietly indexes your pages (the index lives inside the document and updates as you edit). You get:

• the “Related by meaning” section in Searching, and
• the Related palette — pages whose content resembles the current one, with similarity bars. Old notes resurface exactly when they're relevant again.

Privacy, precisely

Everything happens against 127.0.0.1. Encrypted pages are excluded from retrieval and indexing; in password-protected documents the semantic index is itself encrypted. If Ollama isn't running, every AI surface degrades gracefully and the rest of MojoPad is unaffected.