Prototype 01 / Wearable HID
Ambient navigation
device
device
Fused wearable combining machine vision, bone conduction audio, two-way AI voice interaction, and geospatial sensors for continuous, unobtrusive environmental awareness.
Machine vision
Bone conduction
AI voice interaction
GPS + IMU
Two-way query
Contextual AI
Device status
Camera — active scan
Bone conduction — active
AI context engine — ready
GPS lock — acquired
Battery — 84%
Live sensor readings
1.2m
Nearest obstacle
NE
Heading
42dB
Ambient noise
Voice interaction simulator
Ask the device anything — type or speak using the mic button
Bone conduction output
Awaiting voice query...
Focus group discussion prompts
How did continuous environmental narration affect your sense of independence while navigating?
When AI provided uncertain information, how did that impact your trust in the device?
Did foreground/background audio isolation feel natural, or did it cause cognitive overload?
Would this change how you navigate unfamiliar environments versus familiar ones?
Prototype 02 / Smart Cane
Contextual navigation
aid
aid
Evolution of the traditional white cane — gyroscopic machine vision, Bluetooth earpiece, and multi-pattern haptic feedback for real-time contextual awareness during navigation.
Gyroscopic camera
Bluetooth earpiece
Haptic feedback
Obstacle detection
Full blindness support
Haptic feedback simulator
Tap a pattern to trigger cane vibration output
Haptic event output
No events yet...
Vision scan — live camera
Camera initializing...
Bluetooth earpiece paired
Environment interpretation
Press "Scan scene" to interpret current environment.
Camera feed is processed locally — a single frame is sent to Claude Vision for scene description. No video is stored or transmitted continuously.
Bluetooth audio query
Earpiece output
Awaiting query...
Focus group discussion prompts
How did haptic patterns compare to audio cues in terms of usefulness and interruption?
Did the gyroscopic camera feel stable and reliable during movement, or were there moments of disorientation?
Would you use this as a replacement for your current cane, or only as a supplement?
Which haptic pattern felt most natural and which felt confusing?
Prototype 03 / AAUI Layer
Agentic accessible
user interaction
user interaction
Open-source machine vision layer that analyzes any application screen holistically, re-contextualizes it via generative AI, and executes actions via voice — merging CLI power with GUI accessibility.
Machine vision overlay
Agentic AI execution
Generative re-description
Voice command
Any application
Open source
Screen interpreter
Select an app below or type your own — then run AAUI analysis
Instagram
Gmail
Google Maps
Amazon
Microsoft Word
X / Twitter
Uber
Venmo
Active app: None selected
Select an app above to see its elements
Agentic voice command
Instruct the AI agent to perform actions on your behalf
Agent output
Awaiting command...
AAUI — live computer control
checking...
Speak a task and the AI agent will take over the mouse and keyboard to complete it on screen in real time.
Live agent status
Idle — ready for a task
Give the agent a task — type or speak it
Quick tasks
Safety: Move your mouse to the very top-left corner of the screen at any time to immediately stop the agent. Or click "Stop agent" above.
Focus group discussion prompts
How did whole-screen AI re-description change your understanding of the interface compared to element-by-element reading?
When the agent performed an action on your behalf, did you feel in control or did it feel like loss of agency?
Which app would benefit most from this AAUI layer in your daily life?
What would make you trust the agent enough to let it act without confirmation?
Facilitator / App Simulator
AAUI app demo
Let each participant choose an app they use daily and demonstrate the AAUI layer on it live during the focus group session.
Select participant's app
Instagram
Gmail
Google Maps
Amazon
Microsoft Word
X / Twitter
Uber
Venmo
Active: None selected
Screen elements
Select an app above
Agent command
Run a live voice command for the participant
Agent output
Awaiting command...
Log this interaction
Positive
Negative
Question raised
Neutral
Facilitator / Reaction Logger
Real-time feedback capture
Log participant reactions, intensity ratings, and notes for each prototype during the session.
P1 — Wearable HID
Positive
Negative
Question
Neutral
Intensity
★★★★★
Prompt ratings
P2 — Smart Cane
Positive
Negative
Question
Neutral
Intensity
★★★★★
Prompt ratings
P3 — AAUI Layer
Positive
Negative
Question
Neutral
Intensity
★★★★★
Prompt ratings
Facilitator / Session Summary
Session export
Live stats and exportable plain-text log for your research notes.
0
Total entries
0
Positive
0
Negative
All logged entries
No entries yet. Use the Reaction Logger to capture feedback during the session.
Plain-text export
Click "Generate export" to build your session notes.
Copied to clipboard.