Skip to content
← Help

Running Daisy fully offline

Daisy already does the heavy lifting on your own computer: recording, transcription, and speaker labels all run locally, on a model that ships inside the app or that you downloaded once. The one piece that normally reaches the internet is the summary step. This article covers the ways to keep that local too.

Daisy has its own AI model, and sets it up for you. It needs a capable device: iPhone 15 Pro or later, where it is built in, or a computer with a graphics card carrying at least 6 GB of memory. On a computer, first-run setup selects it and downloads it in the background — the download continues even if you close Daisy — and from then on it runs on your machine and needs no key.

(Aside from the AI step, Daisy goes online for update checks, license activation and periodic license checks, model downloads you start, and any calendar feeds or integrations you set up.)

Pick how summaries are produced

You have four choices for the summary step. With the first two, it stays on your own machine or network; with the third, the transcript goes only where you paste it:

  • Daisy's built-in AI — set up for you during first-run setup, and it runs on this computer with no key. This is the default.
  • Local AI server (advanced) — point Daisy at LM Studio or Ollama running on your own machine or network. Setup details below.
  • Skip and use copy-paste — keep transcripts only, paste them into your own ChatGPT or Claude when you want a summary. No setup, no per-meeting cost, no key.
  • Cloud provider with your own key — Groq, OpenAI, or Anthropic. Best summary quality, but the transcript text (not your audio) is sent to the provider over HTTPS with your key.

Set up a local AI server (advanced)

Daisy works with two local AI servers: LM Studio and Ollama.

  1. Start LM Studio or Ollama and load a chat model. Larger models write better summaries, if your hardware can run them.
  2. Turn on Nerd Mode in Settings → Display. In Settings → Providers, find the lm_studio or ollama row under AI provider and select Configure.
  3. Leave Base URL blank if the server runs on this machine on its usual port: Daisy uses http://localhost:1234/v1 for LM Studio and http://localhost:11434/v1 for Ollama. If you changed the port or the server runs on another machine, enter its address ending in /v1. No API key is needed.
  4. Select Test connection & list models →, pick a model, and select Save.
  5. In Settings → Recordings → AI features, set Provider to LM Studio or Ollama. If no provider was set yet, Daisy has already picked the one you saved.

Don't want to run a local AI server? Use copy-paste.

If you'd rather not run a local AI server or pay for an API key, use copy-paste. It's free:

  1. In Settings → Recordings → AI features, set Provider to None — copy and paste.
  2. Open the meeting after Daisy finishes the transcript.
  3. Open the transcript tab and click Copy. The copied text starts with a ready-to-paste prompt.
  4. Paste it into your own ChatGPT or Claude session, ask for a summary.
  5. Paste the summary back into Daisy.

No setup, no key, no per-meeting cost. See meeting summaries for the paste-back step.