Model and instructions

Pick the model that writes your suggestions, give it your own instructions, and run it again over your library.

Mac: Settings → AI
iPhone and iPad: Settings → AI → Model
On: Mac, iPhone, iPad

Choose a model

Resurf picks a model when you connect a provider, except for Ollama. This model writes suggestions, voice note titles and clean-ups, and generated notes. The Assistant has its own model menu on Mac.

Choose Settings… (⌘,) and click AI in the sidebar.
Click the Model button. Models are grouped by provider.
Pick one. Ollama models are listed last and are checked the first time you pick one.

In Resurf's Settings, tap Model, under AI.
Pick a model. If it needs a check, Resurf tests it first.

If the model can't see images, Resurf uses a vision model from the same provider, or organizes images from their text. On Mac, the line under Model says which.

Custom instructions

Tell suggestions how to write titles, tags, spaces and summaries. You edit them on Mac; they sync through iCloud and apply on iPhone and iPad too. They don't change the Assistant, voice note titles or generated notes.

In Settings → AI, click Custom instructions…
Type in any of General, Title, Tags, Spaces and Summary. Blank fields use the defaults.
Click Done. Your text saves as you type.

Not available on iPhone and iPad.

New instructions apply to new suggestions. To update captures you already have, reprocess your library.

Reprocess your library

Runs suggestions again for everything you've saved, with the current model and instructions. It makes one request to your provider per capture, which may cost usage.

Make sure Suggest as you capture is on in Settings → AI.
Click Reprocess beside Reprocess library…
Click Reprocess to confirm.

Not available on iPhone and iPad.

Existing summaries and keywords are cleared first and come back as each capture is processed. On iPhone and iPad, use Generate missing suggestions instead.

Questions

Why do my photos only get text-based suggestions?

Your model can't see images and your provider has no vision model Resurf can use, so images are organized from the text in them.

Where do I pick the Assistant's model?

In the Assistant on Mac, with the button at the left of the composer.

Will reprocessing cost money?

It can. It makes one request to your provider for each capture.