Every app you own
- “open Chrome”
- “Claude kholo”
- “open the app I make 3D models in”
Your Applications folders are scanned at launch. Nothing is hardcoded; install an app and it can be opened by name, nickname or description.
Voice control for your Mac that acts while you are still talking.
Say one long sentence, like “open terminal, add a new tab and let me type”. SayOpen carries out each part the moment it is clear. It lives in the notch, understands English and Hinglish, and knows every app you have installed.
macOS · English + Hinglish · Speech stays on your Mac
Listening…
Apple’s speech engine turns your voice into text on the Mac itself, a few words at a time.
The sentence is cut at “and”, “then”, “phir”, “aur” and commas. Each part is handled on its own.
Jev, TypeSafe’s model, picks the action and fills in the app, site or text. Exact app names skip this step.
The action runs while you are still saying the next part.
The first app is already open while you say the second thing. If one part cannot be done, like opening an app you do not have, it is skipped with a reason and the rest of the sentence still runs.
No command list to learn. Name the app, give it a nickname or describe it. SayOpen matches what you meant.
Your Applications folders are scanned at launch. Nothing is hardcoded; install an app and it can be opened by name, nickname or description.
Browsers, Terminal and Finder. The tab goes to the app you just named, not whichever was last in front.
One-word sites open as the site itself. Web and YouTube searches open straight to results.
Reads menus and buttons through Accessibility and clicks up to four steps. Delete, reset, quit and log out are never offered.
Spotlight plus your Desktop, Documents and Downloads, matched word by word, so a misheard name still finds Cadence_Pack.pdf.
Notes, reminders, volume, media keys, brightness, dark mode, folders, screenshots, lock screen and Enter.
Say “open Wispr Flow” to start dictating and “stop typing” when you are done. SayOpen starts it in the app you are working in.
While Wispr Flow has the microphone, SayOpen stops listening for commands, so a sentence you dictate is typed, never run. When dictation ends, it picks up again.
Wispr Flow is a separate product and needs its own install. SayOpen is not affiliated with it.
A small black island at the top of your screen shows what SayOpen heard and what it did. A menu bar icon holds settings.
Mic levels and a status dot. Sits above full-screen apps.
Your words appear as you say them.
The app’s icon and the time it opened.
Your last five opens, plus Pause, Log and Settings.
Asleep, it ignores everything except “Hey Mac”, “Hey Computer” or “OK Jev”. It stays awake until you say “bye Mac” or “go to sleep”.
Control + Option + Space turns listening on and off. It can open at login.
Speech is recognised on the Mac itself. The only thing that leaves it is the text of what you said, sent to TypeSafe’s Jev model to decide what you meant.
Jev reads the words and returns an action. No audio is sent.
74 real phrasings, run through Jev and through a hand-written rules engine built before the test. Then 15 phrases Jev was never tuned on.
Jev. The rules engine got 43.
Jev. The rules engine got 7.
Latest run. Across runs Jev scores 63 to 68 of 74. A decision takes about 400ms with Jev and under 10ms with rules, which is why exact app names skip the model.
SayOpen is in early access. Join the waitlist and we will email you when your copy is ready. You will need your own TypeSafe API key.
A small universal app for Apple silicon and Intel Macs.
This early build is not notarised by Apple yet. Open it, then click Open Anyway next to SayOpen in System Settings. Only needed once.
Click the waveform in the menu bar and paste a key from console.typesafe.ai. It is stored in your Keychain. No key ships inside the app.
macOS asks the first time you start listening.
Then say what you want. Control + Option + Space also turns listening on and off.
SayOpen is a Mac app from Basement Protocol that lets you control your Mac by voice. You say what you want in one natural sentence and it opens apps, tabs, websites, notes and files as you speak.
No. Speech is recognised on your Mac by Apple’s speech engine. Only the text of what you said is sent, to TypeSafe’s Jev model at api.typesafe.ai, to decide which action you meant.
It understands English and Hinglish commands such as “notes kholo”, and it splits sentences at “phir” and “aur” as well as “and”, “then” and commas.
With the wake phrase on (the default), SayOpen ignores everything until you say “Hey Mac”, “Hey Computer” or “OK Jev”, and goes back to sleep on “bye Mac” or “go to sleep”. Every clause is also checked for whether you are talking to the computer: in testing it ignored all 8 pieces of background chatter.
Say “open Wispr Flow” to start dictating and “stop typing” to finish. While Wispr Flow is using the microphone, SayOpen ignores what you say, so your dictation is never read as a command.
A Mac, microphone and speech recognition permission, and your own TypeSafe API key from console.typesafe.ai, pasted into SayOpen once from the menu bar.
SayOpen uses your own TypeSafe key. When we measured it, a decision cost about $0.00009, so a spoken sentence costs roughly a tenth of a cent. Exact app names like “open Slack” open instantly without a model call.
The early-access build is not notarised by Apple yet. Open it once, then go to System Settings > Privacy & Security and click Open Anyway next to SayOpen. You only need to do this once.
SayOpen is in early access. Join the waitlist on this page and we will email you when your copy is ready.