Cursor prompts. Code review notes. JSDoc and commits that explain why. The work shifted from typing code to typing intent, and the typing tax moved with it. Recitey turns the explanation you would say out loud into the message you would have typed.
Watch a Cursor prompt and a code review reply get drafted out loud in 46 seconds.
Free. Private. Works on most Windows and macOS (Apple Silicon) machines. Runs offline·~3 GB after first launch
Audio stays on your device.Free tier is fully offline.
Press your hotkey, speak, paste. Choose how much help you want.
Direct transcription in under one second. No cloud step, no delay. Pure speed for quick notes.
“Testing again, it seems like it's not moving at all.”
Removes filler words, fixes punctuation and grammar. Your words, just cleaner.
um so basically we need to like fix the config
“We need to fix the config.”
Better word choice and clearer structure, still in your own voice. Doesn't just clean up.
yeah so the deployment is kind of slow we need it faster
“We should review the deployment pipeline.”
The panel records. The sidebar lets you switch modes without opening anything. Watch each mode in action.
can you give the Q3 deck a quick read before 5?
The panel docks by your tray while you're recording.
Docked at the edge while Recitey is running.
A baseline tone, optional writing samples to learn your voice, and hard rules that always apply.
Sets the baseline tone when no per-app rule applies.
Paste short examples of your writing so transcripts sound like you.
Applied on every dictation, on top of samples and preset.
Cursor still blinking.
Runs on your machine. Polishes a sentence in under two seconds, even when it is rewriting your tone from the ground up.
On the free tier, transcription happens on your machine. Whisper turns your voice into text without sending audio anywhere, and it works fully offline. Close your laptop, fly to Tokyo, dictate the whole flight on airplane mode. Same product.
Pro adds optional cloud modes for speed. Cloud rewrite sends only the rough text to Anthropic Claude Haiku for a faster polish. Cloud transcription, if you turn it on, sends your audio for faster results. Anthropic does not train on API data. Keep both off and nothing leaves your machine. You decide what crosses the wire.
What Recitey hears is the same. What it writes depends on the app you're in.
Hi-comma. Paragraph break. Bullets where you list things.
Say “at Alex” and Recitey writes @alex. No commas Slack doesn't want.
JSDoc structure, not email. @returns gets its proper tag.
Anywhere you have a cursor.
Most writing tools assume you start with a blank page and a clear head. Most days, neither is true.
You explained it perfectly on the call. Then you sit down to write the follow-up and the words refuse to come.
Familiar moments. The five-line Slack reply that took fifteen minutes. The PRD section still rough on Friday because you wrote it on Monday and never had time to come back. The code comment that says TODO: explain six months later.
Recitey starts at the moment you actually have the thought.
Press a key. Talk. The polished version appears at your cursor.Not a transcript. The sentence you wish you'd typed.
The transcription happens on your machine, so by default your audio stays on your device. Free is uncapped because there is no per-call cost when the model is local. Pro adds optional cloud modes, with a 30-day refund.
Most voice tools live in a cloud, bill by the minute, and stop working offline. Recitey runs offline by default.
Linux is on the roadmap. Today, this is for the people who write all day on Windows or macOS.
Same keyboard, more words. The job is now writing intent, not just code. Long prompts. Spec comments. Slack threads explaining what the model should build.
Recitey takes the spoken version and pastes the typed version into the chat that's waiting. The hands don't have to catch up to the brain.
The diff is the what. The PR description is the why. After a long pairing session the description gets terse because the typing tax is real.
Recitey starts at the spoken version of the explanation you would have given on the call. Edits are quick because the structure is already there.
Built-in Windows dictation lacks punctuation. Wispr Flow is Mac-first. Recitey runs on both Windows and macOS (Apple Silicon). The accessibility road has been thin on both.
Recitey punctuates and paragraphs without you thinking about it. Hotkey-driven, never menu hunting. The free tier is uncapped because the model runs on your machine.
On-device dictation is free, uncapped. Cleanup and Rewrite get 2,000 words a week on Free, unlimited on Pro.
No credit card required
Uncapped on-device dictation, plus 2,000 cloud-polish words a week.
Billed annually
$120 billed yearly · save 23%
or $13/month billed monthly
For professionals and power users who write daily.
Need it for your team? hello@recitey.com
No. The transcription is local and uses a tuned engine, but the personality is in the rewrite, the per-app tone, the personal style profile, and the floating widget. Transcription is the smallest part of what Recitey does.
Never. Free tier is fully offline. Pro sends only the rough text to the rewrite engine, never audio. The widget shows a "Network: 0 KB sent" indicator while you record so you can verify.
Recitey is available on Windows and macOS (Apple Silicon). Linux is on the roadmap. Tell us your platform at hello@recitey.com and we'll email you the day it ships.
Yes, on Pro. Define your voice once: banned phrases, preferred contractions, custom dictionary, common acronyms. Recitey follows them on every dictation, no matter the mode.
Yes, on Pro. Beyond the rules you set yourself, Recitey quietly studies your past dictations on your machine and builds a short note on how you actually write, then matches it on every rewrite so the result sounds like you, not a generic assistant. It is built entirely on-device. Nothing is uploaded to create it. You can read it, edit it, or turn it off in Settings.
Still works. Median latency goes from ~160ms to ~900ms on CPU-only machines. Fast mode is unaffected. Clean and Rewrite are slower but still under one second on a modern laptop.
Yes. Cancel from settings, no calls, no friction. Free mode keeps working. Your style profile and dictionary stay on your machine. We do not delete anything when you downgrade.
Voice tools market themselves to developers with a familiar pitch: faster input, less typing strain. But if you've spent the...
The shift to prompt-driven development changed what "good documentation" actually means. When you're explaining intent to Claude...
Marcus was 5000 words into a design doc at 11pm, with two more sections to go, when his dictation tool hit the word cap. Back to...
You are 12 minutes into an 800-word design doc, voice flowing clear, the shape of the argument locked in. Then the tool...
You're writing a design doc at 11pm. Settlement architecture, payment flow, edge cases. Your thoughts are coming faster than...
Marcus works on payment settlement at a Series B fintech in Stockholm. Last Wednesday at 11pm, he was dictating a design doc...
Press a key. Talk. Done.