Fundamental v0.9.2

Data honesty

Data source
Synthetic. The question, claims, sources, and confidence values are hand-authored to demonstrate the substrate — they are not a real study or a real dataset.
Snapshot
Static committed snapshot (src/data/ai-evidence.json). It is not produced by scripts/snapshot-examples.mjs and does not come from any live source.
Live refresh
No live refresh — the data is fixed. Only the field's own read-out (the agent-json panel) updates, from the running simulation.
Not claimed
The field represents confidence, support, and contradiction as weight, cohesion, and charge. It does NOT verify whether any claim is true, and makes no medical or scientific assertion — the confidence numbers are illustrative inputs, not findings.

Fundamental · invisible fields · ai-evidence

An agent reads the field, not the DOM.

Demo · synthetic data · substrate proof

One question. Each claim and each source is a body in an invisible field. Citations are real relationships — support pulls a claim and its sources together (cohesion), a contradicting claim carries opposite polarity (charge), confidence is the claim's mass, and stale sources decay. The field is invisible; its effect shows in the type.

The panel on the right is the same field read as data. It is driven by field.query() through an agent-json projection — the structured, serializable view an agent or tool consumes. No DOM scraping: the evidence graph is a first-class queryable substrate. Toggle the field off and both the type and the live JSON go quiet.

Question

Does the vaccine reduce transmission in the studied population?

snapshot · Jun 28, 2026
  1. supports

    The intervention reduces onward transmission

    82% confidence
    • rct Randomized controlled trial, 41k participants 2025
    • cohort Prospective cohort, national registry linkage 2025
    • meta Meta-analysis of 22 trials 2024
  2. contradicts

    No measurable effect on transmission at population scale

    41% confidence
    • observational Retrospective observational, single site 2021
    • preprint Preprint, not peer-reviewed 2022
  3. qualifies

    Any transmission benefit wanes within six months

    58% confidence
    • cohort Prospective cohort, national registry linkage 2025
    • cohort Six-month serological follow-up 2024