Task map, pilot brief, and data-use record — Worksheets

Download the complete part-01 practice pack (ZIP). Extract it and preserve its folders. All organizations, records, and results in this teaching pack are fictional.

These templates support Chapters 1–5. They are starting documents to adapt, not evidence that a workflow has been tested or approved. Fictional examples in the chapter manuscripts must remain labeled when reused.

1. Task map

Use “ordinary software,” “AI draft,” “human decision,” or “controlled action after approval” as useful starting methods. Several can appear within one job.

2. Pilot brief

Pilot name:
Business task and why it matters:
Eligible cases:
Excluded cases:
Owner:
Reviewer and trained backup:
Current process and baseline:
Expected output:
Approved account and input sources:
Allowed actions:
Actions excluded from the pilot:
Review rubric and critical failures:
Development cases:
Held-back evaluation cases:
Measures of quality, time, cost, and business outcome:
Conditions for moving from offline to limited live work:
Conditions for pausing:
Manual fallback:
Start date:
Decision date or evidence requirement:

3. Data-use record

Purpose:
Product, plan, and relevant configuration:
Terms/settings checked by and on:
Official source links:
Permitted information categories:
Excluded information:
Necessary fields for each input:
Approved documents and versions:
People with access:
Enabled connections and read/write permissions:
Training-use settings and exceptions:
Retention, deletion, and logging settings/limits:
Incident contact and first response:
Review triggers:

4. Reference index

Keep conflicting or superseded documents out of the active pack until the owner resolves their role. Preserve historical records separately where needed.

5. Evaluation record

Use one row per attempt. Store only the case information the evaluation needs, with appropriate access and retention. The bundle includes an empty CSV with these fields.

Case ID; case category; offline or live; tool/configuration; prompt version;
source versions; output reference; review result; critical failure;
correction needed; preparation minutes; generation/interaction minutes;
review minutes; rework/fallback minutes; recording minutes; total minutes;
incremental usage cash cost; reviewer; final outcome; escaped error;
notes and next action

Count time consistently. Do not enter the same review or fallback time in two categories. If waiting overlaps with other work, distinguish elapsed time from staff effort before comparing productivity.

6. Pilot decision

Decision: continue / revise / stop
Task and scope evaluated:
Dates and configurations:
Case counts and selection method:
Ordinary-case results:
Difficult-case results:
Critical failures and response:
Quality before and after human correction:
Net handling-time result:
Capacity released:
Incremental cash costs:
Verified cash benefit, if any:
Setup and maintenance effort:
Customer/staff feedback:
Known limits and unanswered questions:
Reason for the decision:
Approved ongoing scope:
Owner and backup:
Manual fallback:
Next review date and triggers:

7. Fictional reference policy for practice

FICTIONAL PRACTICE DOCUMENT — not a real business policy

Source ID: SERVICE-01
Business: Mesa Equipment Service
Effective date: September 1, 2026
Owner: Fictional business owner

The inspection fee is $45.
It is credited toward a repair approved within 30 days of inspection.
Booking is confirmed only after the coordinator checks availability.
No discount or refund may be promised without manager approval.
This document does not specify a warranty, service area, payment method,
or any appointment availability.

Use the practice cases supplied in Practice-Cases.csv to review behavior against this policy. Expected criteria are editor-written examples, not observed model results. Once a case has been used to tune a prompt, do not count it as an independent held-back evaluation.