I Used to Alt-Tab My Way Through Everything
There's a specific kind of friction I dealt with every morning: coffee in one hand, needing to open Slack, pull up my calendar, check a GitHub notification, and switch back to my editor — all before I'd even sat down properly. Mouse in one hand, keyboard shortcuts half-remembered. Nothing dramatic, just slow and annoying.
LacPointer's voice control is the thing that fixed most of that. Not through magic, but through a shortcut I now use more than almost anything else on my Mac: Cmd+Shift+V.
What Cmd+Shift+V Actually Does
When you press Cmd+Shift+V in LacPointer, the bar switches into live voice mode. You speak, it listens, and then it acts. Not just transcribes — acts.
The difference matters. Dictation tools turn your voice into text and stop there. LacPointer's voice control uses the accessibility APIs on macOS and Windows to actually interact with whatever is on your screen. That means it can:
- Open applications by name
- Click buttons in the active window
- Type text into focused input fields
- Control media — play, pause, skip
- Execute system commands
It's the same underlying bar you open with Option+Space for text queries. Voice mode just adds a microphone and a layer of action on top.
A Morning Workflow That Sounds Almost Too Simple
Here's something I actually do. I press Cmd+Shift+V and say: "Open Slack." Slack opens. Then I say: "Check my Google Calendar for today." LacPointer pulls my events and reads them back. Then: "Create a Notion task — review PR for the auth module, due Friday." Done. Three things in under thirty seconds, hands never really left the keyboard in any meaningful way.
The Notion, Calendar, and Slack integrations are baked in — you connect them once in settings and they're available in every voice session from that point on. No re-authenticating, no switching to a browser to authorize something mid-morning.
The Accessibility API Part Is Doing a Lot of Work
I want to be specific about this because it's what separates LacPointer's voice control from a glorified search bar. On macOS, the Accessibility API exposes the full UI tree of whatever application is in focus — buttons, fields, labels, state. LacPointer reads that tree, figures out what action maps to your spoken instruction, and fires it.
So when you say "click the merge button" while GitHub is open in your browser, it doesn't guess based on screen coordinates. It finds the actual interactive element labeled for merging and clicks it. Same principle on Windows with its accessibility stack.
In practice this means you get reliable behavior across app updates, different window sizes, and different screen resolutions. The button is the button regardless of where it happens to be rendered.
Personas Change How Voice Mode Feels
One thing worth knowing: the active persona's voice and system prompt carry into every voice session automatically. If you've set up a Coding persona with a system prompt tuned for terse, technical answers, that's what you get when you press Cmd+Shift+V. If you switch to the Hype persona before a brainstorming session, the voice in your ear is different, the tone is different, the whole session feels different.
You can clone any of the five built-in personas — Default, Coding, Therapist, Stoic, Hype — and tweak the system prompt to be as specific as you want. I have one called "Morning Standup" that's tuned to give short, bullet-pointed status summaries when I ask it to recap my day.
The six available voices (alloy, verse, ash, sage, coral, cedar) are per-persona too. Pick one that doesn't make you want to turn the volume down after ten minutes.
Where It Gets Useful Beyond the Obvious
Opening apps and checking calendars are the easy examples. Here are a few things I've found genuinely useful that aren't immediately obvious:
Typing into forms without touching the keyboard
If you're filling out a repetitive form — addresses, names, dates — voice input into focused fields is surprisingly fast. Say the field content, LacPointer types it. Move to the next field however you like and repeat. Not a replacement for normal typing, but for forms specifically it cuts a lot of back-and-forth.
Controlling media across apps
I keep Spotify running in the background most of the day. Saying "pause music" or "skip this track" without reaching for the keyboard or finding the right window is the kind of small thing that adds up. Media controls work across whatever is currently playing.
Quick GitHub checks
The GitHub skill integrates directly into voice mode. "Any new pull request comments on the auth repo?" gets you a spoken answer. No context switching, no browser tab.
A Few Honest Limitations
Voice control isn't perfect in every situation. In noisy environments the transcription degrades — that's just physics. Very complex multi-step UI interactions that require navigating deep menu hierarchies can occasionally need a nudge. And if an app doesn't expose a clean accessibility tree (some Electron apps are messy about this), the element targeting can be less precise.
None of these are dealbreakers. They're just worth knowing so you set the right expectations. For the common stuff — opening apps, running integrations, media control, typing into fields — it works well.
Try This Right Now
If you have LacPointer installed, press Cmd+Shift+V and say: "What's on my calendar today?"
If you haven't connected Google Calendar yet, it'll prompt you to do it — takes about thirty seconds. Once it's connected, that question becomes a two-second interaction you can run from any context, any app, any point in your day.
That's the actual value here. Not a single impressive demo moment, but a hundred small interactions that used to require tab-switching that now just don't.