How to Dictate Prompts to AI: The Fastest Way to Drive Claude Code, Cursor, and Codex

TL;DR: Whether you're technical or not, your workday is now mostly prompting: Claude Code, Cursor, Codex, chatbots. Typing those prompts is the bottleneck, and speaking is commonly cited as roughly three times faster than typing. Infina lets you dictate prompts to AI anywhere on your Mac, on-device and offline-capable, and it's the only Mac app that closes the whole loop by voice: dictate, say "send it" (or just "enter"), switch to the next agent, repeat. $8.25 a month, billed annually at $99 (at the time of writing) with unlimited words and unlimited voice commands, and a 7-day no-questions refund.

The workflow below assumes the tool can finish the loop. This is what that means. Infina runs on three spoken triggers and there is no fourth. Say "type" and a sentence and it lands as typed text in whatever app is in front of you. Say "send it", or just "enter", and it presses Enter so the prompt actually goes. Say "open Cursor" or "open Slack" and your computer switches apps for you.

Nothing is held down and nothing is pressed. There is no push-to-talk key to find and no Enter key to reach for two hundred times a day, which is the part every other tool in this category leaves to your hands. Every other tool in this category types words into a box and stops there.

Prompting is the new typing, and typing is the bottleneck

Look at what a modern workday actually is. Developers steer coding agents through long natural-language instructions. Writers, marketers, and founders brief chatbots the same way. Nobody is typing code or copy character by character anymore; everyone is typing descriptions of what they want.

Those descriptions are long. A good agent prompt is a paragraph: context, constraint, the edge case you're worried about, what done looks like. Ten of those an hour is a lot of typing.

Speech removes that ceiling. Speaking is commonly cited as roughly three times faster than typing, and unlike typing speed, talking speed doesn't degrade when your hands are tired or your eyes are on another window.

So the math is simple: if prompting is most of your output, and you can produce prompts three times faster by voice, voice prompting is the single biggest speed upgrade available to you. That's the case for learning to dictate prompts to AI, and the rest of this guide is the how.

Why raw dictation is perfect for AI prompts

Here's the detail most people miss: AI models don't need punctuation. They don't need capital letters, tidy commas, or paragraph breaks. A messy, run-on, spoken-word prompt gets you the same quality answer as a manicured one, because large language models read meaning, not formatting.

That flips the usual dictation trade-off on its head. Dictation tools have historically raced to polish output into publishable prose, which takes cloud round-trips. For prompting, that polish buys you nothing.

Infina's base product is built on exactly this insight: raw on-device dictation. Parakeet runs on your Mac's Neural Engine, so transcription is fast, works offline, and your audio never leaves your device. Privacy mode is on by default, so nothing is stored either.

And when you do want polished text (an email, a doc, anything a human reads), the optional $5/month cloud add-on (billed annually at $60) turns on large-language-model cleanup on top of your plan. For prompts, you'll never miss it.

How to dictate prompts to AI on a Mac

Here's the full setup with Infina, from a single dictation to a fully hands-free loop.

Level 1: hold Option and talk

The basic move works in any text field on your Mac:

1. Click into the prompt box (Claude Code's terminal, Cursor's chat, a browser tab, anything).
2. Hold the Option key and speak your prompt.
3. Release. The text appears where your cursor is. Press Enter.

That alone converts every prompt you write from typing speed to talking speed. No per-app plugin, no extension, no setup beyond installing Infina.

Level 2: go hands-free

This is where voice prompting stops being "dictation" and becomes a different way of working:

1. Double-tap the Cmd key to toggle hands-free mode on (it's off by default, and it's our newest surface, still labeled experimental).
2. Speak a sentence that starts with "type", then your prompt. Infina types it into the focused app.
3. Say "send it". Infina presses Enter for you.
4. Say "open Terminal" (or Cursor, or Chrome) and start the next prompt.

Notice what just happened: prompt, send, and app switch, with zero key touches. You can do this leaning back in your chair, two to three feet from the Mac, hands around a coffee. The complete picture of this loop is in our hands-free voice prompting guide.

Level 3: keep several agents busy at once

The hands-free loop compounds when you run more than one agent. A realistic cycle:

• "Type": brief Claude Code on the refactor. "Send."
• "Open Cursor." Dictate the UI change you want. "Send."
• "Open Terminal." Check on the Codex session, dictate the next instruction. "Send."

While each agent works, you review with your eyes and direct with your voice. One person, three agents shipping in parallel, and your keyboard is mostly decorative. This is the workflow the whole product is designed around.

Where it works

Infina types at the OS level, into whatever app has focus. So voice prompting works with:

• Claude Code, in any terminal it runs in.
• Cursor, both the chat panel and inline prompts.
• Codex and any other CLI agent, because a terminal is just a text field.
• ChatGPT, Claude, Gemini in the browser, plus Slack, email, docs, and literally any text field on the Mac.

There is no per-app integration to install or maintain. If you can place a cursor in it, you can speak prompts to it. We've written dedicated setups for voice typing for Claude Code and voice typing for Cursor.

What it costs

Infina is $8.25 a month, billed annually at $99 (at the time of writing), with every update included and a 7-day no-questions money-back guarantee. The $5/month cloud add-on (polish, more languages) is optional and has its own 7-day trial. Details on pricing.

The pricing is deliberately boring: one flat annual figure, unlimited words and unlimited voice commands. No per-seat charge, no per-minute or per-word charge, and no usage meter ticking while you brief three agents in an afternoon.

Requirements: a Mac with Apple Silicon for the on-device models, and English for the base product (the cloud add-on brings more languages).

FAQ

Is it really faster to dictate prompts to AI than to type them? For most people, yes. Speaking is commonly cited as roughly three times faster than typing, and prompts are long natural-language text, the exact kind of writing where dictation shines. The best test is a day of your own prompts, which is what the 7-day refund window is for.

Don't AI prompts need punctuation and formatting? No. Large language models read meaning, not formatting, so a raw spoken prompt performs like a typed one. That's why raw on-device dictation is the right default for prompting, with polish available as an optional add-on for human-facing writing.

Does dictation work inside Claude Code, Cursor, and Codex? Yes. Infina types into whatever app is focused at the OS level, so it works in any terminal, editor, chat panel, or browser tab with no per-app extension.

Can I send a prompt and switch apps without touching the keyboard? Yes, that's Infina's hands-free mode: speak a sentence that starts with "type" to dictate, say "send it" to submit, and "open [app]" to move to your next agent. Toggle it with a double-tap of the Cmd key.

Does dictating prompts require an internet connection? No. By default Infina transcribes entirely on your Mac's Neural Engine, so dictation works offline and your audio never leaves your device. Cloud processing exists only as an optional paid add-on.

Do I need to be a developer to benefit from voice prompting? No. Anyone who spends the day instructing chatbots, drafting with AI, or briefing agents is bottlenecked by typing. If your prompts are longer than a sentence, your voice is the faster input.

The bottom line

The people getting the most out of AI right now aren't better typists. They give their tools more instruction, more context, and more iterations per hour, and the cheapest way to multiply all three is to stop typing prompts and start speaking them.

Start with hold-Option dictation and you're faster today. Graduate to the hands-free loop and you're running multiple agents from three feet away.

A flat $8.25 a month, billed annually at $99, on-device by default, 7 days to get your money back if we're wrong. Your prompts are the work now; say them.

---

Picking the tool that fits this workflow

The workflow above assumes a tool that can finish the loop. Here is how the field lines up against it.

None of these is the wrong answer for everybody. What differs is which reader each one is right for.

The part that is actually different: zero keypresses

The mainstream dictation apps (Wispr Flow, Superwhisper, MacWhisper, VoiceInk, Apple Dictation) are all push-to-talk: you hold or press a key to start talking, and you press Enter to send. That is two physical acts on every single prompt. A hundred prompts in a day is two hundred keypresses that exist only to start and stop the tool. Infina removes both of them. Nothing is held, nothing is pressed, and the same loop runs on three spoken words.

There are three triggers and there is no fourth. "type" types, "send it" sends, "open" opens and switches apps. The send trigger answers to either wording, "send it" or a plain "enter", whichever you find yourself saying. That is the entire vocabulary: no command language, no grammar, no custom scripts to write, nothing to memorize beyond those three words.

When Infina is not the answer. If you write in a language other than English, the base product is English only, and Wispr Flow ships 100+ languages.

Everywhere else, the difference is the loop. Every other tool in that table types words into a box and stops, leaving your hands to press Enter and pick the next window. Infina finishes it: say "type" and your sentence, say "send it" (or just "enter"), say "open Cursor", and keep going without a key.