Tosarah@northwind.co
SubjectRe: Q3 budget draft
Tabfinishes your reply in a professional tone
Native for macOS · Apple Silicon
Type less.
Type less.
Mean more.
TypeNext is one writing layer for your whole Mac. Autocomplete that finishes the sentence, voice that types when you talk, and snippets for the lines you send every day, right where your cursor already is.
- Tab to finish
- fn to talk
- Runs on your Mac
- Free, no account
- Works in every app
on your way?
fn
hold fn to talk · it types it clean
Tabfinish the prompt, get a sharper answer
Tabdraft posts in half the keystrokes
Maya drop your standup when you get a sec
;standupexpands to your saved update
1:1 with Priya
- Ship the redesign by Friday
-
picks up the names and dates on your screen
›
⌥1rephrase the line without leaving the terminal
All in one app
Autocomplete, voice, snippets, and read-aloud.
Reach for whichever is fastest, finish a line, talk it out, drop a saved one, or listen back, without switching apps or breaking flow.
Autocomplete
Tab to finish the line you started.
A faint suggestion appears as you type. Take the next word, the whole line, or a different wording, then keep moving.
Mail · reply to Alex
Auto-playing, press a shortcut to try it yourself.
Knows what's on your screen
Switch on screen context per app, and suggestions pick up the names, threads, and the actual ask in front of you.
Sarahthe login redirect is broken again 😩
On it, I'll take a look.
Writes in your tone
Set a style per kind of app. The same intent comes out formal for email, loose for friends, or short for the team.
Formal
Thanks for flagging this, I'll have a fix out by end of day.
Every suggestion is generated by a model running on your Mac. Your drafts never leave the device.
Voice to text
Speak naturally, AI Auto Edits.
Hold fn and talk casually, half-sentences, filler, rambling and all. TypeNext transcribes, tidies, and drops finished text right at your cursor.
0:00
Hold fn to talk · double-tap for hands-free
Catches your corrections
Change your mind mid-sentence, it writes what you landed on, not the false start.
“let's meet at 2, no wait, make it 3”
Let's meet at 3.
Turns speech into lists
Say it as one run-on sentence and get clean, formatted bullets.
“grab milk, eggs, and bread, oh, and call the plumber”
- Milk
- Eggs
- Bread
- Call the plumber
Speaks 90+ languages
Dictate in your language, transcription detects it and writes it back.
Can you send me the deck before the meeting?
Choose how much to clean up
From word-for-word to a tight rewrite, you choose how much it touches.
um yeah I think we can probably ship it tomorrow, you know, if QA is happy
Snippets
Save what you type over and over.
Write it once, then type a short code like ;sig to paste it. Dates and times fill in for you.
Pick one, watch the trigger expand.
Live variables
Snippets fill themselves in as they expand.
{date} today's date
{time} the time now
{clipboard} what you copied
{selection} selected text
{cursor} where the caret lands
{date:MM/dd} any format
Yours to shape
A full library, fully under your control.
- Trigger by typing (
;cal) or by voice (“my booking link”). - Create, edit, search, import and export everything.
- Leave a
{cursor}mark and the caret lands there, ready to type.
Dictionary
Build your own dictionary.
Add the names and terms you use. Autocomplete and voice typing always spell them right, so autocorrect can’t get them wrong.
NaveenQdrantCaltrainLumakubectlSaaSAnthropic
Watch autocorrect get your words wrong, then your dictionary set them right.
Emoji
An emoji picker at your cursor.
Type your trigger and a proper emoji picker opens inline. Filter by shortcode, move with the arrow keys, hit ↵ to insert, frequently-used up top, skin tone your way.
- Type
:, or set your own trigger, to open it in any app. - Filter as you type; ↑↓←→ to pick, ↵ to insert.
- Choose a default skin tone; your most-used stay on top.
great work everyone
Frequently used
↑ ↓ ← → pick⇥ insertesc
Read aloud
Hear any text in a natural voice.
Select text in any app and press ⌥⌘S. A floating player reads it back in a natural voice, proofreading a draft, catching up on an article, resting your eyes.
- 47 natural voices across 8 languages.
- Runs fully on-device, no cloud, no account.
- Set the speed; pause, resume, or stop from the player.
The launch is on track for Friday. The team closed the last two blockers this morning, and QA has signed off on the build.
Bella · US English
These are real TypeNext voices, tap play to hear them. The app ships 47 across 8 languages.
Per-app control
Different settings for every app.
Give each app its own tone and let it read the screen for context only where you want, then switch it off completely where you don't.
- Per-app tone: formal for email, loose for chats.
- Screen context stays opt-in, app by app.
- Block any app, password managers and terminals stay untouched.
✉️ MailFormal
💬 SlackConcise
📨 MessagesCasual
🔑 1PasswordBlocked
⌨️ TerminalBlocked
Privacy
Your writing stays on your Mac.
Nothing about how you write becomes someone else's training data.
Runs on your Mac.
Autocomplete, read-aloud, snippets and your dictionary, all on-device.
Never trained on. Never sold.
No raw text is logged, synced, or sent anywhere by TypeNext.
Voice is the one exception.
Audio goes to OpenAI with your key, only when you hold fn.
Setup
Up and running in three steps.
-
1
Download & open
Drag it to Applications, launch from the menu bar. No account.
-
2
Grant access, pick a model
Allow Accessibility, then download a local model, the default fits most Macs.
-
3
Start writing
A faint suggestion appears. Tab to take it, hold fn to talk.
FAQ
Common questions.
Is my writing really private?
Autocomplete runs on a local model on your Mac, so your drafts never leave the device, and TypeNext logs no raw text. The one exception is voice dictation, which sends audio to OpenAI using your own API key. If you never use voice, nothing about your writing leaves the Mac.
Does it work in the apps I already use?
Yes · TypeNext works in most Mac text fields: Mail, Messages, Slack, Notes, Obsidian, your browser, code editors and more. You can pause it per-app or per-field whenever you want.
Do I need an internet connection?
Not for autocomplete, snippets, emoji or read-aloud. Those are fully on-device. You'll need internet once to download a model, and any time you use voice dictation (which calls OpenAI).
What does voice dictation cost?
TypeNext itself doesn't charge for it, but it uses OpenAI's transcription API with your own key, so you pay OpenAI directly for what you use. Transcription is inexpensive, and you stay in control of the key.
Can I turn it off for certain apps?
Yes. Every app can be configured on its own, give it a tone, decide whether it may read the screen for context, or block it entirely. Password managers, terminals, and anything else you'd rather keep untouched stay off.
What languages can I dictate in?
Voice dictation understands 90+ languages and auto-detects which one you're speaking. That part runs on OpenAI's transcription. Read-aloud uses on-device voices covering 8 languages. Autocomplete works best in English today.
Which Mac do I need?
macOS 14 (Sonoma) or later, on an Apple Silicon Mac (M1 or newer). The local model loads into memory while it's active, so 16 GB of RAM gives the smoothest experience, though smaller models run comfortably on 8 GB.
How big is the download?
The app is small. The local model is downloaded separately on first run, the recommended default is a few gigabytes, and you can choose a smaller or larger one anytime in Settings.
Will it change my words?
Only when you ask. Autocomplete just offers a continuation. Nothing changes until you press Tab. Keep typing and the suggestion quietly steps aside.
Type less. Mean more.
One native writing layer for your Mac, autocomplete, voice, snippets, emoji and read-aloud.