What if an assistant felt less like an app?
Most AI interactions still begin by opening a screen, finding an interface and explicitly starting a task. A wearable could shorten that distance—but only if it earns a place in daily life.
This project investigates the hardware, interaction and privacy questions behind a compact personal assistant. The direction is intentionally open while early prototypes reveal which moments are genuinely improved by a dedicated device.
Useful moments
Identify small, repeated situations where hands-free assistance has a clear advantage.
Wearable hardware
Explore a physical format that is comfortable, discreet and understandable.
Low-friction input
Test simple controls and feedback that do not require a conventional app screen.
Visible state
Make recording, processing and connectivity states easy to recognize.
A provisional interaction loop.
The architecture below is a working model for exploration, not a finalized implementation. It separates sensing, local device behaviour, AI services and feedback so each trade-off can be tested independently.
Exploration areas
- Embedded systems
- Audio
- AI APIs
- Power
- Industrial design
- Privacy
Principles before features.
Obvious when active
People nearby and the wearer should never have to guess whether sensing is active.
Fast to dismiss
The interaction should end instantly and without navigating through menus.
Focused utility
A small set of reliable jobs is more useful than a broad but unpredictable promise.
Phone optional
The core moment should not depend on taking out another device.
What the next prototype needs to answer.
The priority is evidence, not polish: which interaction deserves dedicated hardware, what feedback feels natural, and what level of battery life makes the object viable.
Interaction rig
Test input and feedback patterns before committing to a compact enclosure.
Measured trade-offs
Add verified details on components, latency, power and user testing as the project develops.