The most useful AI companion is also the most sensitive one. If it can react to your presence, webcam, gestures, voice, focus patterns, foreground apps, and desktop context, then privacy cannot be an afterthought.
That is why AIPET is built around a local-first promise. Sensitive recognition and AI processing happen on your Mac, while online services are limited to product operations such as purchase, license activation, downloads, and updates.
Why desktop AI is different
A chatbot only sees what you type. A true desktop companion can understand richer context: whether you are present, whether focus is slipping, whether a stranger is near the desk, whether you are speaking, and whether the system itself needs to slow down.
That extra awareness is exactly what makes the product useful. It is also what makes a cloud-first design feel wrong for many users.
The AIPET privacy stance
Webcam and face signals are processed locally. Voice and Gemma 3 chat work without a required API key. Focus and activity insights are stored on your Mac. Webcam frames are not uploaded for AI processing.
Local AI, local presence, local control
AIPET uses local models, local computer vision, and local focus data so the companion can react where your work happens. Local-first does not mean the app never connects: activation and software delivery still need AIPET's servers.
Who this is for
This page is for people who are excited by ambient AI, but skeptical of always-online products that collect more than they need. Developers, researchers, writers, consultants, creators, and privacy-sensitive users should not have to choose between useful context and basic trust.
AIPET starts with macOS early access because a careful first desktop release matters. The long-term product direction is broader than one operating system.