Commune Oracle

Collective intelligence

Seven people. One continuously learning visual memory.

Take adaptive quizzes, resolve uncertain details, and score generated images. The Oracle turns every response into structured evidence.

Profile constellation

Health shows what is established and what would be most useful to verify next.

Recent evidence

The newest observations entering the system.

How it learns

Raw observations remain available beneath every conclusion.

1

Ask the highest-value question

Gaps and low-confidence traits come first.

2

Consolidate the evidence

Repeated answers strengthen confidence.

3

Score the resulting image

Visual failures become new quiz targets.

Loading

Preparing the next useful question…

Or answer in your own words The AI will preserve your exact answer and extract every useful profile detail from it.

How confident are you that this accurately describes the person? This controls how much weight the answer receives. Choose “Certain” only for firsthand knowledge.

0 answers this session

Profile journal

Add something important about yourself that no question has asked yet.

Add an image for review

Uploaded images stay in your private R2 bucket and are served through this app.

Select an uploaded imageYour newest uploads will appear in the gallery below.

Teach the Oracle

Rate the selected image across the metrics that matter.

No image selected

Image gallery

Select any image to inspect and grade it.

Improved generation prompt

Built from the selected person’s profile and accumulated corrections. No image API tokens used.

Choose a person and build a prompt.

Evidence by person

Answers, traits, and image scores combined.

System health

What the current dataset can support.

Participation

Who has contributed quiz answers to the shared profiles.

Consolidated traits

The prompt-ready layer derived from raw observations.

Person	Category	Trait	Current value	Confidence	Evidence

The Commune, understood.