type less in every thread
Autocomplete for Slack & chat on Mac
Most of your day in Slack is short replies, status updates, and acknowledgements you type a hundred times. MacAutocomplete predicts the rest of each sentence as greyed ghost text in the message box, so a quick reply takes one keystroke instead of a full line.
thanks team, I'll review and circle back today.
Faster DMs and replies
You start a message, the rest appears inline, and ⇥ Tab accepts it. No popup, no list that steals focus — just greyed text sitting at your caret in the normal Slack composer:
Sounds good — I'll take a look this afternoon and follow up in the thread.
Press ⇥ Tab or → to take the whole suggestion, then send. Keep typing and it disappears — the accept key is only intercepted while a live suggestion is showing, so Tab still moves between fields and Slack shortcuts behave exactly as before. It works the same in DMs, channels, and thread replies, because it hooks the macOS text field rather than the app.
It reads the conversation for context
A good chat suggestion has to know what you're replying to. MacAutocomplete uses the macOS Accessibility API — the same interface VoiceOver uses — to read the text around your cursor in the composer: what you've typed so far, and the message context visible in the field. That's enough to predict a reply that actually fits the thread instead of a generic phrase. It only reads while you're actively typing, it never logs what it sees, and it never touches password or secure-input fields.
Because it lives in the system text layer, the same behavior follows you into every chat tool, not just Slack — see how it works in every app, and how it feels for longer email writing.
Local — your messages never leave the Mac
Every prediction is computed by a model running on your own Apple Silicon Mac. Nothing you type in Slack is sent anywhere — there is zero outbound network for completions, which you can verify yourself with Little Snitch. Work chat is exactly where that matters most: customer names, internal plans, and half-finished thoughts stay on your machine.
| In Slack & chat apps | What you get |
|---|---|
| Suggestion length | Full sentences, not single words |
| Where it works | DMs, channels, thread replies |
| Context | Read from the composer, on-device |
| Network for completions | None — fully offline |
Apple's built-in prediction stops at the next word and only in some apps; this predicts the whole reply, in the chat tools you actually use all day.
Frequently asked questions
Does autocomplete work in the Slack desktop app on Mac?
Yes. MacAutocomplete works in the Slack desktop app for macOS — DMs, channels, and thread replies all get ghost-text predictions in the composer. It hooks the macOS text field, so you don't install a Slack plugin or change any Slack settings.
Does it read my Slack conversation for context?
It reads the text around your cursor in the composer — what you've typed and the message context visible in the field — using the macOS Accessibility API, so suggestions fit the thread. Everything is computed locally on your Mac, nothing is logged, and secure or password fields are never read.
Does the Slack autocomplete send my messages to a server?
No. Every prediction runs on a local MLX model on your Apple Silicon Mac, with zero outbound network for completions — you can confirm it with Little Snitch. Your Slack messages never leave the machine, which is why it's safe for work chat.
Does it work in Discord and Messages too, or only Slack?
Both. Because it lives in the system text layer rather than a single app, the same ghost-text autocomplete follows you into Discord, Messages, and most other chat tools and browsers. Google Docs and canvas-based editors are the main dead zones.
get it
$125 — one-time, lifetime.
Pre-order — ships in ~1 week · lifetime · 3 Macs · runs on your Mac.
Lifetime · 3 devices · 14-day refund (95%).