From software to physical systems

Aero is the software we are building today. Our longer-term research considers how its memory and permissions could support wearable, spatial, and neural interfaces.

Product in development

Personal AI

Aero keeps conversations across model switches. Users can edit saved memory, and Aero can select relevant records for each request.

Approved actions can turn a request into a task or reminder. We are building this layer around clear permissions so future interfaces can use the same context with the user’s consent.

Aero's Today view with upcoming tasks and meetings
Aero interface with fictional example information.

Aero devices

Explore the screened Locket, the slimmer Aero Glasses and Aero Signal's non-invasive EEG fit research. The original voice bench and Desk Dock are available below.

Three-quarter CAD rendering of the Aero Locket with a silver cover, strap opening, and mute slider
Aero Locket mechanical enclosure.

Mechanical fit prototype

Original Aero Locket

A compact voice interface for requests like setting a reminder while your hands are busy.

Enclosure
32 × 44 × 10.4 mm
Cover clearance
0.20 mm per side
Strap opening
4 mm diameter

A metallic face surrounds a recessed request button, a recording light, and an independent mute control. Two separate acoustic ports sit near the lower edge.

Mechanical fit prototype

Aero Desk Dock

A one-piece cradle for setting down the Locket between uses.

Footprint
74 × 62 mm
Height
13.8 mm
Cradle depth
4.9 mm
Cradle clearance
0.60 mm per side

Select Seat Locket in the 3D view to see the original voice enclosure in its cradle. The screened Locket requires a different dock.

Mounting studies

Aero Clip and Loop

A removable clip anchors at the strap opening. A separate split ring lets the Locket hang from a keychain. Remove either mount before seating the Locket in the Dock.

Clip tongue
11.5 × 30 mm
Clip clearance
4.0 mm nominal
Ring envelope
14 mm outer diameter
Download mount studies

Spatial interface concept

Aero Glasses

A frame study for a spatial interface that could show a task or remote workspace while leaving the user's hands free.

Frame width
Approximately 145 mm
Lens envelope
54 × 33 mm each
Components
Frame, temples, optical housing

The model reserves space for optics and an outward-facing indicator. Optical performance, field of view, and comfort remain research questions.

Download GLB

Deliberate voice capture

The wearable design puts a physical mute and a visible recording indicator on the device. A real implementation would also need clear authorization, retention controls, and consent for recording others.

Physical validation

The Locket and Dock geometry is ready for a fit print. Print tolerances, electronics, acoustic performance, retention, and water resistance need physical testing.

View the original wearable sketch

Research direction

Spatial interfaces and robotics

Smart glasses could give Aero a spatial view, and a world model could track relevant objects, relationships, and changes. A separate, authorized control path could connect that understanding to compatible machines.

One research scenario is a person at home using glasses and a voice interface to observe a remote workspace and supervise a robot. The person would see the proposed action, authorize it, and retain the ability to stop it.

Spatial display and human review gate connected to a robot arm within an isometric workcell
Conceptual workcell with a review step between the operator view and the proposed action.

Laboratories

Observe an experiment or manipulate compatible instruments from another location within a defined operating envelope.

Factories

Inspect equipment, monitor a process, or supervise a specific manipulation that can tolerate the connection’s delay.

Accessibility

Explore alternative interfaces for people who could benefit from operating a compatible machine or environment at a distance.

Reliable teleoperation depends on perception, latency management, precise control, safe stopping, and hardware integration. Each task would require validation in its operating environment.

Long-term exploration

Neural interfaces

We want to investigate whether a deliberately learned neural signal could help a person select an action in software or operate a compatible assistive device.

  1. Human intention
  2. Neural interface
  3. Intelligent interpretation
  4. Authorized action
Theoretical interaction pathway for a bounded digital or robotic action.

Non-invasive interfaces would be an initial area to explore. Invasive approaches would require regulated research partners, ethics oversight, and explicit participant consent.

Potential uses include accessibility, assistive technology, and specialized workflows. Each would need evidence for reliability, safety, and the ability to withdraw consent.

Current work

Explore Aero and the prototypes and manuscripts from our adaptive world model research.

Search Rexzen

15 pages found