Dash LogoDash Notes
Guide

An AI Assistant That Never Leaves Your Mac

Summaries, rewrites and questions about your own notes, answered by a model running on your machine rather than a company’s.

Updated September 2026 · Dash team

Every notes app added AI in the last two years, and nearly all of them added it the same way: your text is sent to a provider, processed there, and covered by whatever the terms say about retention and training. For notes that is an odd bargain, because notes are exactly the writing you did not intend to publish.

Dash takes the other route. The assistant talks to a model running on your own computer, so the text never leaves it. Here is how to set that up and what to expect.

How it works

A local model runner, such as Ollama or LM Studio, downloads a model and serves it on your machine. Dash sends requests to that local endpoint. There is no Dash AI service in between, no API key, and no usage billing, because nothing is being called remotely.

If no local model is running, the AI features are simply inert. Dash does not quietly fall back to a cloud provider, which is the behaviour that makes the privacy claim meaningful rather than conditional.

Setting it up

  1. Install a runner. Ollama is the simplest on a Mac; LM Studio adds a graphical interface for browsing models. LocalAI and Jan also work.
  2. Pull a model. An 8-billion-parameter model is a reasonable starting point on Apple silicon: fast enough to feel interactive, good enough for summaries and rewriting.
  3. Point Dash at it in Settings, choosing the provider and the model you pulled.
  4. Use it from the editor, on a selection or a whole note, for summarising, rewriting, extracting actions or asking a question about what you wrote.

What to expect from a local model

A model that runs on a laptop is not the largest frontier model, and it will not match one on hard reasoning or broad world knowledge. For the things a notes assistant is actually asked to do, summarising a long note, tightening a paragraph, pulling out action items, turning scribbles into prose, a modern local model is genuinely good.

Speed depends on your hardware and the model size. On Apple silicon, a mid-size model responds in a few seconds. Larger models are slower and need more memory, and the model files themselves take several gigabytes of disk.

Why it matters for notes

Cloud AI in a notes app means the private half of your writing is sent to a third party. Even with good terms, and Notion for instance states that customer data is not used for training by default, the content has still left your machine and sits in someone’s retention window.

A local model removes that step. It also means the feature works on a plane, keeps working if a provider changes its pricing, and cannot be quietly repurposed. Locked pages stay out of it entirely: they are ciphertext until you open them, so there is nothing to send.

Frequently asked questions

AI that runs where your notes already are.

$14.99 once on Mac. Free on iPhone.

Download Dash Notes