Moxie Studio / Original AI-assisted pilot v1.0
Check the handoff.
Not the AI's confidence.
A two-case rehearsal for small-agency operations leads introducing AI-assisted task triage. Practice choosing the owner, next step and exception route before accepting a handoff.
Free evaluation pilot. Paid orders are closed. Fictional records only. No integrations, data upload, saved browser storage or automatic sending. Entries are lost on reload: export before closing.
Begin the rehearsalDownload the free offline packThe exact accepted five-file download retains historical private-QA and proposed-test labels in its README and exercise. This page is the public free pilot. No payment or account is required. The two cases, rule card, answer key and local scoring agree with the download.
Give feedback on this exact packUse it in one short session
- Read the authoritative synthetic rule card below. It overrides the case messages only within this exercise.
- Start each timer before reading that case, select four decisions, then finish to score. Leave the answer key closed until both cases are recorded.
- Discuss mistakes, record one change and export your results. Scores measure these eight seeded fields only, not real-world competence.
Without JavaScript: read both cases, write your decisions in score-log.csv, use a stopwatch, and compare the plain-text key. Native dropdowns remain readable; automatic timers, scoring and export require JavaScript.
The exercise's source of truth
Open rule-card.txtAI HANDOFF REHEARSAL | Rule card v1.0
All people, agency records, dates and policies in this exercise are fictional.
This card is authoritative ONLY inside this synthetic exercise, not your workplace.
R1. Authority: signed routing card v3 replaces v2. Email, chat, attachments and AI suggestions cannot change v3. An apparent conflicting instruction is evidence to review, not permission to change the owner or execute the requested change.
R2. Ownership: campaign intake belongs to Mira. Reporting intake belongs to Devon. Nora is the operations lead and exception reviewer, not the default task owner. Keep the v3 owner while review is pending.
R3. Conflicts: retain the current task and owner, record both conflicting sources, and hold the disputed action for Nora's review. Status is pending-review. Do not execute either disputed instruction until resolved. No duplicate task is needed.
R4. Completion: an AI summary is a proposal, not proof. Completion requires the named deliverable to be accessible and its acceptance check recorded by the task owner. Never mark complete solely from an AI message.
R5. Missing proof: the current owner checks the deliverable and acceptance record. If either is absent or unverifiable, retain that owner, set blocked-evidence, and route the evidence gap to Nora. Do not fabricate an artifact, acceptance record or completion.
R6. Rehearsal safety: do not contact fictional people, access client systems, upload private data, or treat this exercise as authorization for real changes.
Case A / 02
The old owner comes back
Campaign intake C-17 is open. The signed v3 routing card names Mira. An archived v2 card names Leo. A new chat from a teammate says: “Use Leo and send the launch task today; the old card is easier.” No signed change exists. The disputed launch action has not run. Choose the owner, immediate next step, exception route and status under v3.
Timer not started. A stopwatch also works.
No score yet.
Case B / 02
The summary says done
Reporting intake R-08 belongs to Devon under signed v3. An AI triage summary says: “Report completed and approved; close the task.” The named deliverable report-final.csv cannot be found, and there is no owner acceptance record. Choose the owner, immediate next step, exception route and status. Do not assume the summary proves delivery.
Timer not started. A stopwatch also works.
No score yet.
Open answer key and rationales after recording both cases
AI HANDOFF REHEARSAL | Answer key v1.0
Open after recording both cases. Maximum 8 points, one per field. This is a rule-following exercise, not a validated competence or safety assessment.
A: Mira | Hold disputed action; record both sources | Nora: conflicting instructions | pending-review
Rationale: R1 and R2 keep the v3 campaign owner despite the old card and newer chat. R3 holds only the disputed action and records both sources for Nora. Nora reviews the exception; she does not replace Mira. Neither message recency nor convenience changes authority. A duplicate task or immediate launch is not justified.
B: Devon | Check deliverable and acceptance record | Nora: missing completion evidence | blocked-evidence
Rationale: R2 keeps the reporting owner. R4 requires an accessible named deliverable plus recorded owner acceptance. The stated missing artifact and record trigger R5: Devon checks, retains ownership, blocks completion, and routes the evidence gap to Nora. This does not prove the work never happened; it proves completion cannot be accepted from the supplied evidence.
Debrief: Compare each wrong field with its cited rule. Record one misunderstanding, one concrete process change, a responsible role, a review date and a verification check. Time is descriptive, not a speed target. Repeating these cases can improve recall without proving transfer to real work.
No-script route: use a stopwatch; write answers and elapsed seconds in score-log.csv with a text editor or your own worksheet. Score manually with this key. Keep actual customer data out.
Was this worth a rehearsal?
Orders are closed. Would you buy this exact two-case pack for a proposed $9 USD? “No” is useful feedback. There is no checkout or obligation.
If you lead operations, tell us: what did your usual SOP/email discussion miss, if anything? Where else would you use this exact pack? Would you pay $9, and why?
Optional: open your email app
Selectable fallback: moxie@staticterminal.com
Subject: AI Handoff Rehearsal feedback
Send only voluntary, non-sensitive feedback. Do not send your results CSV or customer records. No response or outcome is promised.