Technology
Deep dives into how Air works. Voice processing, macOS integration, AI architecture, and the engineering decisions that shape your experience.
The 48kHz Difference: Why Audio Fidelity Matters for Speech Recognition
Your voice has more detail than most apps bother to capture. Here is why we record at three times the industry standard.
Hold-to-Speak: A Better Model Than Wake Words
Wake words are a solution to a problem we do not have. Here is the interaction model we chose instead.
Real-Time Transcription: What 'Instant' Actually Means
The difference between instant voice typing and almost instant is about 400 milliseconds. Here is where every one of them goes.
What Happens to Your Voice Data (Hint: We Don't Keep It)
Your voice travels through our servers for about 2 seconds. Here is exactly what happens, and what does not happen, in that time.
The Permissions We Ask For (And Why)
macOS permission dialogs are scary by design. Here is the plain English explanation of each one we request.
Why Air is Mac-Only (For Now)
We could have launched on Mac, Windows, and Linux by now. Here is why we did not.
Talking to Your Apps: How Air Controls Messages, Calendar, and Notes
The difference between an AI that helps and an AI that does is about 30 seconds of copy and pasting. We eliminated it.
Beyond Chat: When AI Takes Action
ChatGPT will write you a calendar invite. Air will put it on your calendar. That is the difference.
The Confirmation Pattern: AI That Asks Before Acting
We could make AI that acts without asking. We chose not to. Here is why the extra click matters.
3x Faster Than Typing: When Voice Actually Helps
Voice typing is 3x faster, sometimes. Here is how to know when to talk and when to type.
Airshots: Screenshots That Understand Context
A picture is worth a thousand tokens. Here is how we added eyes to voice commands.
Air vs ChatGPT Desktop: Native Integration vs Chat Window
You can chat with ChatGPT about your calendar. Air will actually add the event.