Transcribe the spoken source
The AI transcription capability turns supported speech, recordings, or voice notes into text. The transcript becomes the source for later classification.
Turn a voice note into the task it describes.
Ghost can transcribe supported audio, separate several spoken intents, extract stated dates and context, and propose the right work object before anything is saved or scheduled.
- Direct answer
- Voice to task is a workflow that transcribes a voice note, identifies one or more actionable intents, proposes structured task fields, and preserves the transcript so the user can review wording, dates, and destination before saving the work.
One complete example.
A voice note says, “Send Maya the revised brief Thursday, save the pricing idea, and check the launch page Friday.” Ghost separates the task, note, and possible scheduled check, shows the extracted dates, and asks the user to confirm each destination.
Where each capability enters.
This is one workflow across Ghost, not a bundle of disconnected products. Each feature owns a specific stage and keeps its own access boundary.
Create a trackable action
A confirmed action can become a Focus item with stated timing, priority, source transcript, Kanban state, and Epic context.
Keep future execution explicit
A spoken request to run work later may become a scheduled AI task after the time, timezone, recurrence, and brief are confirmed. That is different from a simple to-do.
How the work moves.
The handoff stays inspectable from source to result. Every stage names what it received, what changed, and what needs review.
Receive supported audio
The user supplies a voice note, recording, or supported channel input. Availability and quality depend on the audio path and configured speech services.
Transcribe before interpreting
Ghost produces text that can be reviewed. Names, dates, and domain terms deserve particular attention because a confident task built on a bad transcript is still wrong.
Separate multiple intents
A spoken brain dump may contain a note, task, question, reminder-like request, and Calendar action. The workflow routes them separately instead of forcing the entire recording into one object.
Extract stated details and ambiguity
Ghost proposes the action, owner, date, recurrence, priority, and source context it can support. “Next Friday” or an unnamed person should be confirmed when the intended meaning is not clear.
Review the destination
The user confirms whether the result belongs in Focus, Notes, Scheduler, or a supported Calendar operation through Integrations. Saving one does not silently perform the others.
Keep the transcript available
The resulting item retains the transcript or source identity so later work can return to the original wording and correct the structured record if needed.
What the workflow does not assume.
Connected work still needs accurate sources, explicit destinations, and review where an action changes another person’s system.
- Transcription quality varies with audio, names, accents, noise, and the configured input path.
- A spoken date should be confirmed when its timezone, recurrence, or intended object is ambiguous.
- A task, Calendar event, note, and scheduled run have different effects and should not be created interchangeably.
Questions about voice to task.
Can Ghost turn a voice note into several tasks?
Yes. After transcription, Ghost can separate several actionable intents and propose individual Focus items. The user can review the wording, timing, and destination before saving them.
Can a voice note create a reminder?
Ghost can route a future action into the appropriate supported task or scheduled-work flow after timing and recurrence are clear. A Focus timing label is not automatically an alarm or notification.
Can Ghost create a Calendar event from speech?
A supported Calendar action can be proposed through Integrations, but event creation is separate from saving a to-do and should be confirmed because it can affect time and attendees.
Does the task keep the voice transcript?
The workflow is designed to retain the transcript or source identity so the structured action can be checked against the original spoken request.
What if Ghost hears a date or name incorrectly?
The transcript and extracted fields should be reviewed, especially when names, dates, timezones, or recurrence affect other people or future execution.
Start with one real workflow.
Bring the source, the intended result, and the boundaries that matter. Ghost is in private beta for macOS.
Request access