How to Navigate Continuous Close

Use this page as the runbook for the dense Continuous Close runtime surface. The goal is not to read every card. The goal is to prove one story: Finance can know where the books stand right now, why, what blocks certification, what action is required, and how the decision can be replayed.

Open live demo cockpit Start the 4-screen runtime story Open introduction Run finance as of now

Do not navigate randomly

The page has too much information if used as a dashboard tour. Use it as a proof theater: executive position, blocker, case, evidence, policy, action, replay, and credibility.

Primary storyClose now

Primary proofReplay

Primary personaCFO + Controller

Primary riskCertification blockers

Six-click demo path

1Run finance as of nowOpen with the current financial position. Ask: if Finance closed right now, what would the books look like? 2Run live cockpitUse guided buttons to inject the event, open evidence, approve action, and replay. 3Open the 4-screen storyUse the runtime journey section. Do not scroll the whole page first. 4Drill into blockersShow which mission, country, entity, and owner prevent certification. 5Prove lineageShow source systems, generated timestamp, calculation basis, agent, confidence, and provenance. 6Replay the decisionReconstruct signal, context, policy, evidence, human gate, action, and learning.

What each user should look at

Audience	Question	Where to click
CFO asks	Are we ready to close and what changed?	Start with readiness, confidence, exceptions, expected variance, and why-not-100%.
Controller asks	Can I certify this?	Open blocker details, evidence, control outcome, owner, age, and certification boundary.
Technology asks	Is it real?	Open credibility payload, source lineage, generated timestamps, agent trace, provenance badges, and replay.
Analyst asks	What do I do now?	Open the case drawer and use recommended action, policy, evidence, and escalation path.

1. Executive Landing

Use this only to frame the question: readiness, confidence, expected variance, open risks, and why not 100%.

Say: “This is not a close dashboard; this is a current financial position.”
Click readiness only after explaining the CFO question.
Avoid reading every metric aloud.

2. Mission / Blocker View

Move from enterprise readiness to the specific blocker that prevents certification.

Show mission readiness and owner.
Show materiality, age, and SLA risk.
Use this to prove actionability.

3. Analyst / Controller Case

Open one case and show the heavy lifting: evidence, policy, journal preview, recommendation, and human gate.

Say: “Agent does the investigation; controller owns judgment.”
Show approval boundary before action.
Do not over-position autonomy.

4. Runtime Proof

This is for architects and skeptical stakeholders. It proves the number is not an unexplained tile.

Show source lineage and timestamp.
Show agent contract and confidence.
Show provenance badges for seeded baseline vs live query.

5. Replay

Use replay as the trust moment. It answers how the runtime reached the recommendation.

Replay signal to decision.
Show policy and evidence used.
Show human approval or escalation.

6. Integrity / Controls

End with enterprise safety: fail-closed SOR boundary, evidence coverage, fallback, and control status.

Say: “AI enhances the runtime; it does not own the control boundary.”
Show missing SOR credentials are disclosed, not hidden.
Show no uncontrolled write-back.

12-minute talk track

Minute	Action	Line
0-2	Open current position	If Finance closed right now, this is readiness, confidence, expected variance, and what blocks certification.
2-4	Open the 4-screen story	We use one event to move from CFO view to mission control to analyst action to runtime proof.
4-7	Open one case	The runtime assembles evidence, applies policy, recommends action, and keeps controller judgment explicit.
7-9	Open credibility	Every key number has source, timestamp, calculation method, owner, agent, and provenance.
9-11	Open replay	This is how audit reconstructs what happened and why.
11-12	Open integrity	The production SOR boundary is fail-closed until production credentials exist. No overclaiming.

What not to do

Do not scroll top-to-bottom and explain every section.
Do not lead with agents. Lead with current financial position and certification risk.
Do not call a seeded validation baseline production data. Use the provenance badges.
Do not imply autonomous posting without policy, evidence, approval, and SOR gateway.
Do not stay on the UI only if architects ask if it is real. Move to CLI/runtime proof.