Adaptive colour intelligence · v0.4

Adaptive colour and
design intelligence.

THE MAZE learns from visible response patterns to predict colour and spatial design preferences — quietly, in the background.

Start Design Review
How it works

Four quiet steps from looking to learning.

01

Experience colours or design images

Smooth gradient sessions or cinematic image slideshows. No tasks. No questions during the session.

02

Camera observes visible response locally

Attention, calm, engagement and discomfort signals are processed on-device. No raw video leaves your machine.

03

Reinforcement learning adapts in real time

A lightweight policy reweights what to show next based on observed response — exploring, then refining.

04

A calm report is generated

After enough trials, results appear: predicted favourite colour, ranked designs, and a confidence estimate.

Privacy first

Your face stays on your machine.

THE MAZE is a research instrument, not a surveillance tool. Camera processing exists only inside the active session window. Nothing is uploaded.

Read the privacy notice
No raw video is stored
Only derived numerical signals are held in memory during the session.
No identity recognition
No face matching, no face embeddings, no biometric profile.
Local processing only
Models run in your browser. The session ends when the tab closes.
Results are estimates
Not a medical or psychological diagnosis. A measurement of visible response.
Use cases

Where THE MAZE earns its keep.

INDIVIDUAL

Personal colour discovery

A calm 5-minute session that surfaces the colour family you actually respond to.

STUDIO

Interior design review

Present three palette options to a client and walk away with a defensible direction.

PRACTICE

Architecture client presentations

Compare schemes for a renovation — courtyard, atrium, screened porch — under one report.

ACADEMIC

Research and affective computing

A reproducible setup for quantifying visible response to controlled visual stimuli.