Outcome 01
Shared intuition
Product managers recognize the gaps a requirement leaves open and know which ones produce defects no check can catch.
ObservableGiven a requirement, learners name the decision it avoids before proposing a solution.
Diagnose what a requirement fails to decide, write criteria that carry their own evidence, bound the scope, survive independent ratification, then hold the product claim to the evidence that exists.
Stage 9 needs a facilitator-supplied outcome signal the room has not seen. Stage 8 is strongest run jointly with engineers on course 301, who escalate the same policy gap from the same fixture.
Learning outcomes
The stages succeed when practitioners recognize the same risks, describe them in the same words and change what they do before approving work.
Outcome 01
Product managers recognize the gaps a requirement leaves open and know which ones produce defects no check can catch.
ObservableGiven a requirement, learners name the decision it avoids before proposing a solution.
Outcome 02
Product and engineering describe criteria, evidence types, limits and unresolved policy with the same words.
ObservableLearners state what a clean evaluation run does and does not support, using the same evidence types the engineers use.
Outcome 03
Before work starts, product managers write criteria with evaluation paths, bound the scope and submit the spec for independent ratification.
ObservableLearners produce a ratifiable specification for an unfamiliar outcome without the stage worksheets.
The learning path
Each stage adds one decision and produces an artifact the next stage works from. Preparation happens before instructional time.
Sibling repository: ../payments-validation-fixture. Complete this before the first stage.
Read the product spec and the outcome signal, then decide what the requirement does not settle, ranked by the cost of shipping without an answer.
Rewrite the intent as an outcome hypothesis: the signal, the affected users and the business condition that should change, without naming a mechanism.
Turn the hypothesis into acceptance criteria where each one states observable behavior and carries an evaluation path.
State what this slice does not do, so scope creep becomes a visible specification decision rather than an invisible delivery one.
Write the specification so it decomposes into four to eight file-scoped tasks, and split it when it does not.
Submit the specification for independent ratification, then revise it until it is accepted or every objection has an owner and a date.
Read the Round 0 evidence bundle the engineers read, and state what it lets you claim to a customer.
Take the F8 escalation, make the policy decision, and state what it changes downstream.
Take an outcome signal you have not seen and produce a ratifiable specification, without the stage worksheets.
Eight classes of AI mistake and the six types of evidence
Current versus V1
The target sequence is explicit about what exists now, what the new stages require and where delivery is still blocked.
The current fixture exposes diagnose, plan, validate and report. validate executes the workflow and report summarizes its evidence. Results may appear before Stage 2 method selection.
diagnose → review → eval → define → plan → run. Review methods are chosen before results are shown; define writes defects/*.md; plan writes tasks.json; run creates the next diff. Stage 7 is blocked until the unseen assessment change exists.
Where this sits
Tracks at the same number are peers on the same fixture. The 501 assumes both of them.