I used to hate switching windows just to send a message
Picture this: I'm deep in a file, following a train of thought, and I suddenly need to ping someone on Slack or check tomorrow's calendar. The old workflow was painful — Alt+Tab, find the window, click into the input, type, go back, lose the thread. Every single time.
LacPointer's floating bar already cut half of that friction when I started using Option+Space to ask questions without leaving my editor. But the voice mode — Cmd+Shift+V — cut the other half entirely. Now I just talk.
What Cmd+Shift+V actually does
Hold Cmd+Shift+V anywhere on your desktop and LacPointer drops into voice mode. You speak, it listens, and then it acts — using macOS and Windows accessibility APIs to actually operate your machine. Not just generating text for you to copy. Actually doing the thing.
That distinction matters. A lot of "voice assistants" produce text output you still have to act on. LacPointer's voice control can:
- Open applications by name ("open Finder", "launch VS Code")
- Click UI buttons by describing them ("click the Send button", "press Accept")
- Type text into whatever field is focused ("type my standup update")
- Control media playback — pause, skip, volume — without touching the keyboard
- Send Slack messages, create Notion tasks, check your calendar — through the integrations layer
It's all wired through the accessibility APIs, which means it works across native apps, not just websites. On macOS, it hooks into the Accessibility framework. On Windows 10/11 64-bit, it uses the equivalent UI Automation APIs.
Real examples from my actual day
Here's how voice control shows up in my workflow, not in theory but in practice:
Morning standup
Before I open my laptop properly, I hit Cmd+Shift+V and say "create a Notion task: review PR for the auth refactor, due today". Done. The task is in Notion before I've touched the trackpad. The Notion integration handles the actual API call; voice just feeds the intent.
Slack without switching tabs
Mid-coding, I say "send a Slack message to the dev channel: I'm pushing the fix now". LacPointer sends it. I never left my editor. The conversation thread stays intact in my head.
Media while reading docs
This one sounds small but I use it constantly — "pause music", "next track", "volume down" without reaching for the keyboard. Especially useful when I'm reading something dense and don't want to break my scroll position to fumble with media keys.
Opening apps from voice
Some mornings I boot up and just say "open Terminal", "open Figma", "open Safari" one by one while I'm still getting coffee. Minor thing, but it stacks up.
How it knows what to do
Each persona in LacPointer has its own voice setting — alloy, verse, ash, sage, coral, or cedar — and its own system prompt. That system prompt is active during voice sessions too. So if you're running the Coding persona, the voice mode responds with terse, direct answers. If you switch to Therapist, the same voice session becomes slower and more reflective.
The active persona's system prompt and voice bake into every session automatically via the ephemeral token, so you don't have to reconfigure anything when you switch between them. It just works however you've set the persona up.
If you haven't played with personas yet, it's worth a look — you can build one from scratch under LacPointer settings with a custom name, avatar, voice, and system prompt. The voice mode will carry all of it.
What it doesn't do (yet)
I want to be straight here. Voice control is not magic. It works best for things LacPointer has direct integration with — Notion, Slack, GitHub, Calendar — and for OS-level actions like opening apps and controlling media. If you're trying to get it to interact with a very obscure third-party app that uses non-standard UI components, it might miss the target. The accessibility APIs have blind spots.
Also, it's not a replacement for a full voice-to-code workflow. If you want AI to write and execute code in your terminal from voice commands, that's more lac-cli territory — specifically lac shell and lac agent for the heavy lifting.
Voice control inside LacPointer is best thought of as a hands-free layer for the integrations and OS actions you'd normally do by switching windows.
Setup takes about 30 seconds
If you haven't enabled it yet: open LacPointer with Option+Space, go to settings, and make sure microphone access is granted. On macOS you'll get the system prompt the first time you hit Cmd+Shift+V. On Windows it's the same — first use triggers the permission dialog.
After that, it's live. No further config unless you want to tune the persona it runs under.
One practical tip
Build a persona specifically for voice use. Call it something like "Hands Free", give it a short and punchy system prompt like "You control my PC and integrations by voice. Be brief. Confirm actions in one sentence. Do not explain what you're about to do, just do it." Pick a voice that feels right (I use ash for this one — it's calm and doesn't feel robotic), and set it as your default when you're in a flow state that you don't want to break.
It completely changes how you interact with your machine when both hands are otherwise busy.
Download LacPointer at lacai.io/download — available for macOS and Windows.