The working A prototype: use its controls to test the build question.
Explain one defensible idea to different audiences without changing the underlying truth.
RN conceived the question, structured the evidence boundaries, designed the experience, implemented the prototype and documented its limits.
This independent prototype demonstrates an approach; it is not a deployed client system, professional advice or proof of real-world outcomes.
Work through the controls with a situation of your own rather than the sample values. The tool responds to what you put in, so the useful output comes from real input, and everything is processed in your browser as you go. Any sample content you find already loaded is there to show the shape of a filled-in state, and you can clear it and start again at any point.
The build overview sets out the question, the purpose, the role and the evidence status in one place, and the public record behind it states what changed, what another person can reuse and what supports the system. If you want to judge how far this build should be trusted, the record is the page to read, not this one.
BUILD 002-A
ACTIVE PROTOTYPE
ACTIVE PROTOTYPE
Multi-Audience Meaning Architecture
Change the language for the person in front of you without quietly changing the claim underneath it.
If you have one idea that needs to make sense to different people, you can use this. You do not need communications training, prompt-engineering knowledge, or linguistic terminology.
The adaptation layer and the truth-check layer are separate. This prototype transforms register locally in the browser, then audits numbers, negation, uncertainty, named anchors, and facts you explicitly lock. Later model-based generation can plug into the same verification architecture without becoming the judge of its own output.
01 / SOURCE TRUTH
What must stay true?
02 / DOOR
Who needs to understand it?
general public
not runPrioritize what it means and why it matters. Avoid specialist shorthand.
03 / CLAIM GRAPH
1 propositionBreak the source into things that can drift.
A human reviewer may reduce risk only if they can see enough evidence, intervene before the outcome, and actually change or stop the decision.
- numbers
- —
- negation
- —
- uncertainty
- may
- names
- —
04 / ADAPTATION
Different door. Same factual core.
engine / deterministic local100low drift risk
In plain language: A human reviewer may reduce risk only if they can see enough evidence, intervene before the outcome, and change or stop the decision.
✓ Numbers preservedNo numeric anchors
✓ Negation preservedNo negation anchors
✓ Uncertainty preservedmay
✓ Named anchors preservedNo named anchors
✓ Locked facts preservedNo locked facts
✓ Concept coverage93% of source content terms remain verbatim
✓ Expansion controlled8% net word expansion
05 / LANGUAGE
Translate on the device when the browser can.
not run
Translation is another transformation layer, not proof of cultural equivalence. Consequential multilingual communication still requires competent human review.
SOURCE
A human reviewer may reduce risk only if they can see enough evidence, intervene before the outcome, and actually change or stop the decision.
GENERAL PUBLIC
In plain language: A human reviewer may reduce risk only if they can see enough evidence, intervene before the outcome, and change or stop the decision.
06 / PORTABLE RECORD
Take the reasoning with you.
The share state contains the source, audience, style, locks, and target language—not a server-side account. The JSON export adds the claim graph, audit and local provenance events.
LOCAL PROVENANCE
No optional model/translation actions yet.
What this prototype does: separates generation from verification. Deterministic checks remain independent even when a supported desktop Chrome browser supplies a local language model or translator. Passing automated checks is not proof of semantic, legal, scientific, clinical, or cultural equivalence.
Change the door. Do not move the room behind it.
A high preservation score is not proof that two statements mean exactly the same thing. For legal, scientific, clinical, regulatory, multilingual, or culturally sensitive communication, review the transformed version with the relevant human expertise.