Speak your prompts to any coding agent.

Try it with your next prompt

1. Install Entune with the command below. You need no account or speech API key to start.
2. Download a local speech model in Models, or paste a key for a cloud one.
3. Set a shortcut, click into your agent’s prompt, hold the shortcut and say what you want built.

Puts Entune in Applications and opens it. The first start can take up to a minute.

Entune pastes into the text input you are in, whether that is an agent’s chat box in your editor or a prompt in your browser. On a Mac, delivery into a terminal has been tested in Ghostty. Wherever Entune finds no input to paste into, the text stays on the clipboard and its pill says so. Every transcript is also kept in History.

Say everything, not the short version

A good prompt carries the goal, the constraints and the cases to watch for. Speaking makes it easy to include all of it instead of trimming the request to save typing.

• Long prompts: Entune itself sets no recording limit; how long a recording each speech model handles is up to that model. With AssemblyAI, recordings of over 20 minutes have been transcribed.
• Fast mode: while you speak, Entune transcribes each part at a natural pause, so when you stop, only the last part is left. It works with every speech model and stays off until you turn it on.
• Structure: turn on paragraphs and bullets, and a decision model adds breaks and list items where your dictation clearly has them, keeping every word. Filler removal takes out um, uh and other hesitation sounds in English. Both are off until you turn them on.

Keep your project’s words right

Library names, flags, acronyms and the agents’ own names are where speech models slip. Entune learns a dictionary from your own recordings for the speech model you dictate with, and when a word could be your term or an everyday one, a decision model picks by the sentence, so ordinary words stay as you said them. Dictating to Claude Code shows how, with measured results.

Let your agent add the words you confirm

Any agent that can run a command can add a correction to your dictionary through Entune’s local API, once you have confirmed it. Put the instruction in the file your agent reads, such as CLAUDE.md or AGENTS.md. The ready-to-use instruction asks before it adds anything and sends only what you confirm. A confirmed word replaces its sound wherever Entune hears it, so keep this for sounds you never mean literally. The local API has no password: use Entune on a computer only you use.

Your code stays yours

Prompts often describe code you would rather not share. A local speech model keeps your audio on your computer. If you turn on corrections, filler removal or formatting with a cloud decision model, that provider receives the text it works on; Laya keeps it on your computer, in English. The data and privacy reference lists exactly what leaves your computer.