HOW IT WORKS · INFORMATIONAL PREVIEW

From context.
To a reviewable plan.

See how a changed fact can reopen a plan without losing what mattered—or the reasoning that came before.

YOUR CONTEXT

What matters shapes
what comes next.

A new timetable changes the picture. The wish to learn stays; the next step needs reconsidering. Follow one question through the intended APU → Circle → RealityFutures → Agentz relationship.

Fictional, non-financial illustration. All records and responses are authored examples—not live app outputs, a model run or advice based on your worksheet.

Can an evening course fit around existing commitments?

APU · Purpose and limits
Make room to learn without displacing shared commitments. Keep Wednesday 18:00–20:00 protected.
Circle · Known in this example
The authored draft timetable lists Monday 18:00–20:00. The second session is not listed.
RealityFutures · Possibilities
The missing session could fit or clash. No probability has been estimated; missing information is not a 50% forecast.
Agentz · Conditional strategy
Given the declared priority to protect shared time, first check the complete timetable. Record fit, clash or unknown, then stop for review. No message, booking or payment.
Value and cost
The learner spends attention to keep the learning option open without committing shared time. Deferring avoids that effort but leaves the question unresolved. The timetable does not choose these priorities.
Alternatives and affected people
Defer or compare another course. People sharing those commitments would need to discuss any change to shared time.
If the evidence changes
In the authored revision, a Wednesday 18:30–20:30 session overlaps the protected time. Prefer checking one alternative course, then return with fit, clash or unknown. This costs search effort and may delay learning; no suitable alternative is established. The earlier plan needs reconsideration; enrolment remains undecided.
Review remains separate
Writing down a need for agreement does not obtain that agreement or create permission to act.

Read the example here even when the interactive module is unavailable. Checking, pausing and seeking review are all valid endings.

Frame your own question →

The worksheet starts blank. This illustration does not fill it in or evaluate your answers.

ONE CAPABILITY · DISTINCT EXPERIENCES

Agentz across the apps.
A dedicated home here.

The intended experience connects the context you permit, what is known, what could happen and the next step you want reviewed. Each app keeps its own sources and access rules. Changed evidence should reopen affected plans—not silently rewrite them or act for you.

Agentz.cloud is the dedicated destination for the family-office experience and visible Murmur guideline concept. This website does not demonstrate those cross-app connections. The local planning worksheet is deterministic formatting, not a Soul Loop run.

THE SOUL LOOP

Explore. Audit. Stage.
Never self-authorize.

L1 — CHECK THE INPUT

Untrusted, stale, contradictory or unreceipted input first receives a bounded defense check. The mission must fit what direct evidence can resolve.

L2 — EXPLORE ALTERNATIVES

Develop genuinely different possibilities. Circle's sourced observations and RealityFutures' forecasts retain their own uncertainty and provenance.

L3 — AUDIT THE OPTIONS

Compare evidence, assumptions, risks and uncertainty. Agreement between agents is not independent proof.

L4 — STAGE THE PROPOSAL

Make the smallest defensible next step reviewable: what supports it, what remains assumed, and when to stop or seek further review. Refusing or escalating can be the right endpoint. In development dispatch, L4 alone makes scoped local changes. That rule does not establish a deployed execution service; the product’s outward proposal stays unsigned.

L5–L7 — ESCALATE WHEN NEEDED

Architecture, negative-boundary and constitutional questions receive non-deployable counsel, then return to L4. Counsel adds no authority.

FEEDBACK — TEST BEFORE ADOPTION

Independent outcome evidence may seed a change proposal. Owner adoption and a later scoped L4 act are required before changing the system.

The default L2 → L3 → L4 route applies only after trusted, exact-scope L1 intake clearance. Otherwise, L1 comes first. This describes the routing contract; it does not claim a live integration or autonomous service.

THE AUTHORITY BOUNDARY

A recommendation
is not permission.

UNSIGNED BY DESIGN

Agentz's terminal artifact is an unsigned proposal. It must not sign, authorize or borrow authority from an audit, model consensus or a visual status badge.

EXTERNAL AUTHORITY

Actual mandates, recognized persons, scope, thresholds, expiry and revocation belong outside the reasoning ladder. A wallet signature alone does not establish every applicable mandate.

A strategy needs
accountable follow-through.

The intended experience goes beyond writing a proposal. It keeps the decision, the task and its result connected—without treating them as the same event.

Fictional continuation of the course example. These are separate intended states, not a live task runner, approvals or completed actions. This optional explanation is not a checklist required for every everyday decision.

Inspect the six distinct states

  1. Propose a bounded task.

    Compare one alternative course’s public timetable. The proposed task owner is the learner. Spend at most 20 minutes; return with fit, clash or unknown. No enrolment, contact or payment.

  2. Decide who may do what.

    The learner decides whether to do that check. Any delegation needs its own explicit scope and permission. Changing shared commitments requires the affected people’s agreement; listing it does not obtain it.

  3. Do only the authorized task.

    The responsible person or separately authorized execution service carries it out within the agreed limits. Stop after one comparison or 20 minutes, whichever comes first. Agentz does not create the mandate.

  4. Record what happened.

    A dated receipt would identify the timetable checked and the task performed. That is evidence of activity, not proof that the course fits or the learner’s life improved.

  5. Check the actual result.

    Check the published timetable and the learner’s permitted constraints independently of the task’s completion claim. Missing sessions leave fit unknown. Completion alone is not success.

  6. Reconsider the next step.

    Use that outcome evidence to propose continuing, changing course or pausing. Keep earlier reasoning visible. A revised proposal requires its own decision; no automatic enrolment follows.

DESIGN BOUNDARIES

Claims stop
where evidence stops.

NO SELF-AUTHORIZATION

Staging does not confer signing or execution authority.

NO WARRANT TRANSFER

An observation, forecast and recommendation do not inherit each other's evidence.

NO FIXTURE AS LIVE PROOF

A local test or illustrative interface cannot establish live grants, deployed execution or independent outcomes.

NO PROMISED EXIT WITHOUT TESTS

Custody, privacy, revocation and recovery claims require implementation-specific evidence before an operational release.

NEXT STEPS

Inspect before you act.

Explore the proposed workflow and its open release gates. This preview asks for no funds, keys or identifying information.

Review release status →