PaulHold a key.
Click where you want to write. Hold Fn to begin, then release when you're done.
fnTHE FIRST MAN WHO REALLY LISTENS. 😉
You talk.
Paul listens.
Finally, a man who listens.
And aims to remember what you actually said.
English and German. Mixed together, too.
More languages to come.
Made for Mac · Apple Silicon · macOS 14+
Current alpha interface. Your voice stays on your Mac.
LESS FRICTION. MORE YOU.
He listens. You get
the last word.
Say it your way.
Speak English, German, or a mix of both. Listen mode has no grammar-rewriting stage.
Keep your words close.
Recognition runs on your Mac. Review recent dictations, copy a result, or confirm a vocabulary correction.
↳ On your MacBUILT WITH CARE. TESTED WITH YOU.
A good listener
keeps learning.
Paul is a working prototype. We're preparing a small beta to test real voices, real apps, and the places where recognition still gets it wrong.
Faithful dictation is the goal. Recognition can still miss or invent words, so review important text.
- Local English & German dictation
- Working prototype
- Fn shortcuts & text insertion
- Working prototype
- Gentle audio preparation
- Optional filter
- Only me voice focus
- Experimental preview
- Polish & light editing
- In development
THE FIRST LISTENERS
Help Paul
listen better.
Meet Paul. Give him an earful.
Different voices. Different Macs. Honest feedback.
A few good listeners first.
Paul is being tested in a small private alpha. We’re improving real-world recognition before opening the public beta.
The beta will be free. No card and no automatic subscription.
Explore the plans ↗Public registration and downloads aren’t open yet. We don’t collect email addresses on this page.
A few things to know.
Does my voice leave my Mac?
The current recognition and voice-profile pipelines run locally. Models are downloaded during setup. The prototype retains recordings locally for review; retention and deletion controls are part of the beta release work.
Can Paul ignore everyone except me?
Only me is an experimental review mode with three voice samples. It isn't yet validated for dependable background-speaker rejection, overlapping speech, music, or TV. Automatic insertion is withheld in this mode.
Can I download Paul now?
Not publicly yet. We're preparing the signed beta package, installation checks and feedback process before inviting testers.
Will Paul rewrite what I say?
Listen mode has no additional grammar rewrite. The recognition model can still normalize or mishear speech. Optional spoken punctuation commands and confirmed word corrections can change the output. A separate Polish mode is planned.