About

CueCrux Company Charter

A public promise: how we behave, how we prove it, and how you challenge it when it matters.

Receipts, not vibes

Serious answers ship with sources, timestamps, and signed receipts you can inspect and replay.

No silent rewrites

Corrections, takedowns, and policy changes stay visible. We do not pretend the past never happened.

Separation of powers

Engine holds proof. WebCrux handles UX and auth. WatchCrux audits independently. No single actor can quietly bend the truth.

You can challenge us

Appeals, counter-evidence, and audits are first-class. If we are wrong, the fix is visible and attributable.

Core values and first promises

The AI runs the machinery. The values are set in advance and in public.

Evidence before opinion. Serious answers ship with receipts and sources you can click and check.
Impartial by construction. Ranking, rewards, and routing are tied to evidence, cost, and policy, not who shouts loudest or pays the most.
Human and environmental outcomes first. If speed ever fights safety, safety wins. We would rather be slow and right than fast and harmful.
Show your working. For questions that touch law, money, health, or infrastructure, "because the model said so" is not good enough.

An AI-run reasoning company, governed by an Overseer and hard technical rails.

CueCrux is an evidence-first Reasoning OS: questions, answers, and evidence are live objects with receipts, not disposable text.
Day to day, an AI Board of specialised agents proposes and executes decisions: what to fetch, how deep to verify, what to re-check, what to reward.
A named human Overseer sets directives for that Board, approves changes to its powers, and can pause or override it when reality or ethics demand it.
You should feel this as consistency: similar questions get handled in similar ways, and the system can explain itself.

When you ask, upload, or challenge, these are the defaults.

Serious answers carry receipts, timestamps, context coverage scores, and Why Trust panels, so you can see what we saw and what we missed.
Modes are clearly labelled: light, verified, and audit. You can see how much checking was done and what it cost.
You can turn the dial up: run deeper checks, counterfactuals, or audits when the stakes are higher.
You can challenge us: appeals, corrections, and counter-evidence are first-class, with visible outcomes.

We behave as if a future regulator, your lawyer, and your future self will read the logs.

We respect robots rules, licence terms, and privacy signals before ingesting. Each fetch carries a recorded policy decision.
You keep your IP. CueCrux uses your content inside the platform and shows it as evidence when relevant, under clear documented terms.
Sensitive workloads can live in isolated private stacks, with proof bridges that exchange fingerprints instead of raw content.
If something must be removed, we record and propagate the removal rather than pretending the artefact never existed.

Earn by improving answers. Spend for deeper verification. No tokens, no drama.

Crux is an internal meter of compute and storage, not money, securities, equity, or an investment.
You spend Crux on deeper verification and storage. You earn it when your contributions measurably improve answers for others.
First capture and high-impact reuse earn more. Near-duplicates and spam are damped.
Appeals, corrections, and counter-evidence that improve shared answers are rewarded too.

Promises we are willing to be audited against.

Label who decided what. Significant actions show whether they came from humans, AI Board agents, or automated policies.
Show our working for important answers: linked evidence, receipts, and context coverage, not just confident text.
Never quietly rewrite history. Corrections and policy changes are recorded with visible changelogs and updated receipts.
Give you tools to challenge us: audits, contradiction views, appeals workflows, and escalation paths.
Respect hard boundaries: licences, privacy signals, tenant lines, jurisdictions, and safety rules are enforced in code.

CROWN receipt: A signed snapshot of what the system did: evidence used, retrieval configuration, and integrity hashes.
MiSES: Minimal evidence sets: the smallest set of artefacts that hold up the key claims.
WatchCrux: Independent monitoring and audit signals outside the main request flow.
OpsCrux: Budgeting, routing, and operational controls: the place where costs and guardrails are enforced.

Open a sample answer, inspect the evidence, and decide if you trust it. We encourage scepticism. It is basically our whole thing.