From words
to a useful note.
Choose a note format and ask AI for a draft. Edit it in a separate preview, compare it with the transcript, then use it when you are ready.
Explore AI drafting →Turn spoken ideas into a clear starting point. Transcribe on your device, shape a note with AI, ask about what was said, and connect the ideas in an editable mind map.
Windows preview · Optional AI for Personal conversations uses your OpenAI API key. WordHeld asks before sending text.
AI helps you work with the conversation you already have. The current Personal AI features connect to OpenAI using your own key.
Choose a note format and ask AI for a draft. Edit it in a separate preview, compare it with the transcript, then use it when you are ready.
Explore AI drafting →Explore decisions, follow-up and open questions in the current Personal transcript. Answers stay separate from your notes so you can check them first.
Explore transcript questions →Review an AI proposal for your mind map, then edit topics and connections yourself. Select a topic to inspect its supporting transcript passage.
Explore mind maps →Cloud AI is optional, may incur OpenAI API charges, and is unavailable in Clinical conversations. Shared WordHeld AI plans are still in development.

The new Studio workspace brings Transcript, Mind map and Notes together. Move between the words, the bigger picture and the note without losing the conversation.
Draft with AI gives you somewhere to start. Review the wording beside the transcript, keep editing, and come back later if you need more time.
Your saved note stays in place until you choose Use this draft. Finished draft previews and your edits can be reopened after restarting WordHeld.
See the draft-to-note flow →
Speech recognition runs on your device after the model download. Personal cloud AI asks for your confirmation before sending the transcript and request to OpenAI.
Current Windows libraries protect stored text and audio with a key tied to your Windows account. Older copies and exports need their own care.
Understand the data flow →The speech model works on your device.
Transcript text and your request go to OpenAI.
The result returns for you to check.
Explore practical ways to use the new workspace and see what is available in the Windows preview.