Use lessons 01–03 or equivalent skills before this exercise. You can revisit them from the lesson list.
A text editor and supplied fictional decision packets. No purchase, procurement action or operational approval is performed.
Start with the idea
Plan a small trial with checks that can stop it.
Define the scope, measurements and responsible decision owner before the trial. Include missing-answer cases. Measure quality and total effort separately.
Stop condition: An observable event that pauses or ends the trial.
See the example and check your understanding
A made-up teaching example
The fictional FAQ pilot uses ten cases, including two unanswered by the sources. A human approves release. Restricted content in output pauses the trial even if routine answers are fast.
Ask yourself
Can faster answers cancel a restricted-content failure?
Show the explanation
No. Apply the stop rule. Keep assignment and sending outside any unapproved automatic behavior.
Start with these checks
- Define source-support and missing-answer checks.
- Count total review effort.
- Name who can pause and resume.
Use the different practice case below and check every requirement in its full instructions.
Full learning objectives
- Choose a baseline and evaluation set.
- Specify review and stopping rules.
- Include affected users and operational constraints.
Captions appear here during playback.
Design a bounded pilot with stop conditions
Plan a small trial with checks that can stop it.
Video playback is unavailable. Read the descriptive transcript below, go directly to practice, or open the video file.
Source-based teaching example · Fictional data · Reviewed 2026-10-03. Microsoft Mark · locally synthesized. On a small screen, use fullscreen for diagram details or read the transcript.
Read the descriptive transcript
0:00 · What you will learn
Plan a small trial with checks that can stop it.
What appears on screen
Plan a small trial with checks that can stop it.
0:05 · The idea
Define the scope, measurements and responsible decision owner before the trial. Include missing-answer cases. Measure quality and total effort separately.
What appears on screen
Define the scope, measurements and responsible decision owner before the trial. Include missing-answer cases. Measure quality and total effort separately.
0:19 · One useful term
Stop condition. An observable event that pauses or ends the trial.
What appears on screen
An observable event that pauses or ends the trial.
0:26 · A worked example
The fictional FAQ pilot uses ten cases, including two unanswered by the sources. A human approves release. Restricted content in output pauses the trial even if routine answers are fast.
What appears on screen
The fictional FAQ pilot uses ten cases, including two unanswered by the sources. A human approves release. Restricted content in output pauses the trial even if routine answers are fast.
0:42 · Your turn to think
Can faster answers cancel a restricted-content failure? Pause and decide before continuing.
What appears on screen
Can faster answers cancel a restricted-content failure? Pause the video if you need more time.
0:59 · Why this answer works
No. Apply the stop rule. Keep assignment and sending outside any unapproved automatic behavior.
What appears on screen
No. Apply the stop rule. Keep assignment and sending outside any unapproved automatic behavior.
1:09 · Check 1
Define source-support and missing-answer checks.
What appears on screen
Define source-support and missing-answer checks.
1:15 · Check 2
Count total review effort.
What appears on screen
Count total review effort.
1:20 · Check 3
Name who can pause and resume.
What appears on screen
Name who can pause and resume.
1:24 · Now try a different case
Open the lesson and try its different practice case. Use every check in the full task. Watching alone does not mark the lesson complete.
What appears on screen
Open the lesson Try the new case Check and revise The full task and its requirements are in the linked lesson.
Watching explains the example. Complete the lesson practice to check what you can do. Explore all visual lessons →
The method
A pilot is a decision experiment with a defined scope, duration, owner and evidence package. Select representative tasks and meaningful negative cases. Measure correctness, total effort, required assistance and failure recovery separately. Explain who reviews outputs and what happens when an acceptance criterion is missed.
Set pause criteria before results are known. A critical unauthorized action can require stopping regardless of average speed. Include accessibility and user workflow constraints in evaluation. Use synthetic data until actual use is authorized. Budget planning can be recorded as assumptions; this exercise neither spends money nor initiates subscriptions.
Choose your model, harness and offline route
Text-only assistant
Use supplied text. Ask for a plan or draft; do not claim browsing, code execution or file access that was unavailable.
Tool-enabled application or harness
Record which tools actually run, allowed files and actions, approval boundaries and observable results. Tools are optional unless the task needs them.
Local or cloud deployment
Inspect inference, retrieval, connectors, logs and storage separately. Use only authorized environments for real data; training fixtures are fictional.
Higher-capability or frontier model
Evaluate the exact task and configuration. Capability does not supply missing evidence or grant permission. No provider ranking or guarantee is implied.
Offline alternative
Use the supplied inputs and inspect the worked example manually. Builder exercises use local deterministic Python. No paid account or live model call is required.
Record the exact product/model label when available, tools actually used, inputs, permitted actions, observed output and checks. Live accounts or APIs may have external costs; they are optional and are not used by this page.
Read the general workflow guide →Worked example
Supplied input
Fictional pilot: draft public FAQs from supplied sources; ten cases include two absent-answer questions. Human reviewer approves release. Stop if restricted content enters output.
A checked result
Measure supported claims, abstention on missing answers, total review effort and release decisions. A failure on restricted content pauses the trial even if other cases are fast.
Why this works
The release gate is tied to a consequential failure rather than an average score alone.
Try a different case
Independent case: meeting-action extraction for a volunteer team. Plan a two-week synthetic trial with normal notes, unassigned owners and disputed decisions. Define baseline, owner, outcomes and pause rules.
Download this task and checklist →Interactive practice / Fictional fixture
Set the pilot’s gates
Use the supplied independent case above. Check the bounded decisions, then review your written artifact against the full acceptance criteria.
Everything runs in this browser. No AI model is called and no work is uploaded. Bounded checks cover specific facts and rules. Your writing is not graded: you review it yourself. Practice never marks a lesson complete automatically.
Define baseline, reviewer, normal/negative cases, measures, pause rules and the final decision.
Try the task, run the checks, then revise. Checks are teaching feedback, not an instructor grade.
Observed browser simulation
This trace comes from the bounded local checker. It is not a Python run, model result or production tool execution.
Keep or restore your work
Drafts are not saved automatically. Save only fictional practice data. A browser draft may be lost if storage is cleared; download a copy to keep it.
Your steps
- Specify who reviews and what constitutes a usable register.
- Include unknown owners and unagreed dates.
- Define the decision to be made after the pilot.
Check your work
- No automatic assignment or sending.
- Quality and total effort are measured separately.
- Pause criteria and responsible decision owner are explicit.
After your attempt: review notes and a common mistake
A practical pilot might require correct preservation of unknown fields and no unauthorized actions. Thresholds must suit the consequence; do not borrow an arbitrary passing percentage from another workflow.
Different wording or a different justified decision can pass. Check your work against every requirement; these notes are not an automated grade.
Explain your approach
How will the pilot avoid rewarding unsupported confidence?
Keep your artifact, one correction or decision, its evidence and one remaining limitation. Nothing on this page sends or uploads your work.
Check your understanding
Choose an answer to see feedback. This is a knowledge check, not instructor assessment.
Review your independent practice
Complete the supplied task before recording progress. Use the acceptance criteria and reveal the review notes after your attempt. Keep your artifact yourself; nothing is uploaded.
Not yet self-reviewed. Completion stays in this browser and does not issue a badge.
Optional deeper study
Explore the related general-course lesson → This opens the existing canonical lesson; it is optional depth, not another copy of this focused task.
Revision 2026-10-03. Fictional examples, not observed provider results. Knowledge checks and completion records are self-study aids.