Skip to content
Transcript.audio

Settings: AI

The AI tab: AI is included in your plan, with no keys to paste. How transcripts and notes are made, presets, models for each task and on-device backup.

Last updated

AI is included in your plan, so there are no keys to paste. The defaults work well for most people.

How should the AI work?

Pick a preset. Each one chooses the best model it can for every job, using only what your plan includes.

PresetWhat it means
Optimal (the default)The best model for every task.
RecommendedThe best balance of accuracy and cost.
Most accurateThe top model for each task.
CheapestThe lowest cost that still gives usable notes.
FastestThe shortest wait after the call ends.
Privacy optimisedAudio and notes stay on this computer. See Privacy optimised mode.
FreeOn this computer only. Notes need Ollama or LM Studio; without them you get the transcript only.
CustomShown when you pick models by hand.

Models for each task

Each job can use its own model: Transcription, Meeting notes, Details to capture, Titles and folder names and Help assistant. Each shows what it is Now using and what it falls back to (If it fails: ...), so if one model is busy the next takes over. Choose a model by hand and the preset becomes Custom; Use preset choice puts it back.

Premium models are part of Plus and Pro. The models page describes each one.

On-device backup

Under Transcription, On-device backup transcribes on this computer when the cloud cannot, and audio never leaves your computer. It downloads about 600 to 700 MB once. Its line shows Ready, Downloading 42%, Waiting to download or Off.

Still stuck? Contact us and we’ll help.