How to Talk to JetBrains AI Assistant by Voice on a Mac
JetBrains AI Assistant lives inside IntelliJ IDEA, PyCharm, WebStorm and the rest of the JetBrains IDEs, but it expects you to type your prompts. There is no microphone button. The fix on a Mac is a system-wide voice to text app that types into the AI Assistant chat box for you, so you can describe what you want out loud instead of tapping it out.
Key takeaways
- AI Assistant accepts typed prompts only, so voice input comes from a separate dictation layer.
- A system-wide app types your speech into the chat box in any JetBrains IDE.
- On-device voice to text keeps your prompts and code local instead of uploading audio.
- AI cleanup and a custom dictionary turn rambling speech into precise, well-punctuated prompts.
Why talk to JetBrains AI Assistant instead of typing?
Prompting an AI coding assistant is really about describing intent: "refactor this function to use async await and add error handling," or "write a unit test for the edge case where the list is empty." That kind of sentence is faster to say than to type. Most people speak around three to four times faster than they type, so long, descriptive prompts are exactly where voice pays off.
Voice also lowers the friction of asking follow-up questions. Instead of stopping to type, you keep your hands near the keyboard for editing and simply speak the next instruction. If you already dictate elsewhere, this is the same habit applied to your IDE. It pairs naturally with the broader practice of coding by voice on a Mac, where dictation drives comments, commit messages and prompts rather than raw syntax.
How voice input reaches the AI Assistant chat box
The key idea is that a good Mac dictation app is not tied to one program. It works system-wide, in any app or text field, which includes the AI Assistant prompt box docked inside your JetBrains IDE. When you trigger dictation, the app listens, transcribes on-device, and inserts the finished text wherever your cursor sits. From the IDE's point of view, it is identical to typing.
The cleanup step matters here. Raw speech is full of "um," "like," and false starts, and it arrives with no punctuation. On-device AI cleanup, powered by Apple Intelligence, strips the filler, fixes the punctuation and grammar, and hands the assistant a tidy prompt. That is the same reason on-device AI can rewrite text without the cloud: the model runs locally on Apple Silicon, so nothing has to be uploaded to polish it.
Set it up in five steps
Install a system-wide dictation app
Download BlaBlaType for macOS and grant the accessibility and microphone permissions it asks for. These let it type into any app, including your IDE.
Pick your shortcut and model
Choose a global shortcut you can reach one-handed and select a local model. The on-device Whisper and Parakeet models run offline, so there is no per-minute cloud cost.
Open AI Assistant in your JetBrains IDE
In IntelliJ, PyCharm or WebStorm, open the AI Assistant tool window and click into its chat input so the text cursor is blinking in the prompt box.
Press the shortcut and speak your prompt
Describe what you want in plain language: the refactor, the test, the bug you are chasing. The transcribed, cleaned text lands in the chat box as if you typed it.
Add tricky terms to your dictionary
Teach the custom dictionary your framework names, variables and internal jargon so they are spelled correctly every time, then send the prompt with Enter.
Keeping your prompts and code private
When you read code aloud or describe a proprietary system, that audio is sensitive. Cloud dictation services upload it to their servers to transcribe. On-device dictation does not: the audio and the transcript never leave your Mac. That is a meaningful line for anyone under an NDA or working in a regulated codebase.
One honest caveat: the dictation layer only controls how your voice is handled. Once the text is in the AI Assistant chat box, JetBrains sends that prompt to whichever model provider is configured, under JetBrains' own privacy terms. If you want a fully local model in the loop as well, some developers pair local runtimes such as Ollama with their editor. The point is to be clear about which step is local and which is not.
| Step | Cloud dictation app | On-device app |
|---|---|---|
| Your spoken audio | Uploaded to a server | Stays on your Mac |
| Transcription | Runs remotely | Runs locally |
| Works offline | No | Yes |
| Types into AI Assistant | Yes | Yes |
| Prompt sent to AI model | By JetBrains | By JetBrains |
Talk to your IDE, privately
Dictate prompts into JetBrains AI Assistant and any other app, with AI cleanup and 100% on-device speech. No card needed for the trial.
Download for macOSTips for better voice prompts
Speak in complete thoughts. AI cleanup will fix grammar, but a clear sentence produces a clearer prompt. When you need punctuation or symbols spelled out, say them, and use custom AI prompts to shape tone or default formatting for the kind of instructions you send most. If English is not the language you think in, remember the app handles 90+ languages with optional translate-as-you-speak, so you can dictate in your native language and have the prompt arrive in English.
The same workflow extends beyond JetBrains. You can use it to talk to ChatGPT with voice on your Mac, dictate into your terminal, or draft documentation. For a wider view of options, see our roundup of the best dictation software for Mac in 2026. And if you are curious how other voice tools handle privacy, OpenAI documents its own approach in the Voice Mode FAQ.
Frequently asked questions
Can I talk to JetBrains AI Assistant by voice on a Mac?
Yes. JetBrains AI Assistant has no built-in microphone button, but you can dictate into its chat box with a system-wide voice to text app. On a Mac, an app like BlaBlaType types your spoken words into the AI Assistant prompt field in any JetBrains IDE, exactly as if you had typed them.
Does voice dictation into JetBrains send my code to the cloud?
It depends on the tool. Cloud dictation apps upload your audio to their servers. On-device tools like BlaBlaType transcribe every word locally on your Mac, so your spoken prompts and any code you read aloud never leave the machine. Note that JetBrains AI Assistant itself may still send your prompt to its model provider once it is typed in.
Will voice to text understand programming terms and library names?
Modern on-device models handle common technical vocabulary well. For unusual names, framework terms or internal jargon, use a custom dictionary so the app spells them the way you want every time. On-device AI cleanup also fixes punctuation and removes filler words automatically.