Modes
Build custom dictation workflows with shortcuts, app triggers, and AI processing.
Modes let you set up dictation differently for each context. Every mode has its own keyboard shortcut, can target specific apps, decides where the result goes, and can run your transcript through AI. Open the Modes section to manage them.
The default mode
The first mode in the list is your default. It runs whenever no other mode matches, so you can dictate anywhere without extra setup. You can change its shortcut and AI instructions, but you cannot delete it or limit it to specific apps. To start over, open it and choose Reset to defaults.
Create a mode
Click New Mode, give it a name (for example Email, Slack, or Coding), then set how it activates and what it does. A mode with no AI instructions simply types out your raw dictation.
Activation
Each mode can be triggered two ways:
- Keyboard shortcut. Pick a preset or record your own, then choose how it behaves: Hold or Toggle (hold to talk, or tap to keep recording), Hold (push to talk), or Toggle (tap to start, tap to stop).
- App triggers. Add one or more apps so the mode activates automatically whenever you dictate in them. App triggers are available for every mode except the default.
AI Instructions
In the AI Instructions field, tell the AI how to transform your dictation, for example "Translate to English" or "Turn this into a polished email". Leave it empty to keep your raw dictation.
Where the result goes
Under Advanced settings, the Result goes to option controls delivery:
| Option | What it does |
|---|---|
| Auto Insert | Types the text into the focused app |
| Paste and Send | Pastes the text and presses Enter |
| Copy to Clipboard | Copies the text without typing |
| History Only | Saves to history without inserting |
Choosing an AI model
AI processing runs on Spokenly's hosted models with Pro, or for free with your own API key. To use your own key, open LLM Providers, add a provider (OpenRouter, OpenAI, Anthropic, and more), and test the connection. You can then select that provider per mode as the Text Model.
With your own key, Advanced settings also expose a Fallback Text Model (used when the main one is unavailable), Reasoning effort, Temperature, and a custom System prompt.
Test a mode
Click Test inside the mode editor to record and see the result without leaving the screen. The result dialog shows both what you said and the AI output.
To clean up transcripts automatically, see Word Replacements. To improve recognition of names and jargon, see Dictionary.