The right ritual for the right question.
Triggers, participants, and outcomes are clear. Use existing rituals. Starting cadences can be adjusted without adding a meeting marathon.
Daily · up to 10 minFlow CheckBuild · Value · Learning
Enable the next item to be completed and surface missing feedback.
Delivery Team; Product Owner and Loop Guide when decisions are needed.
AI supportThe Learning Guide summarizes approved board, test, and review data. The team corrects the current status.
Active work, pending acceptance, specific failures, and the remaining investment budget.
Connection to the loopsB5 · B6 · B8 · B9 · V3 · L1
- 0–2 min
Correct the status
The team checks the AI preparation. Missing entries do not mean missing work. Bug fixes, support, and small tasks belong in the visible workflow.
- 2–7 min
Keep work flowing
What can be finished next? Where is a decision, review, or help needed? Finish waiting work before starting something new.
- 7–10 min
Agree on the next action
Set an owner and the next evidence point. Trigger a Spec Clinic when the budget is tight or the spec is unclear.
Concrete outcome
A concrete next action with an owner, a visible impediment, or a focused clarification task.
Shopping cart example: The stock check passes, but business feedback is missing. A Feature Partner takes the next Acceptance Slot.
Open the related promptWeekly · 45 min as a starting pointRefinementBuild · Value
Clarify the problem, intended impact, and next small change together.
Product Owner, Delivery Team, and Feature Partner. The Measurement Partner joins for measurement questions. The Loop Guide supports when needed.
AI supportScout collects questions and examples. Planner checks dependencies. Impact Analyst prepares measurement questions.
Task and any related strategic initiative, fixed constraints, delegated decision rights, user examples, priority, available work capacity, and a proposed time and AI budget.
Connection to the loopsB1 · B2 · B3 · B4 · V1 · V2
- 0–5 min
Limit the investment
The Product Owner and team check the commissioned boundary and decide how much effort and AI budget to use within their mandate. The appropriate decision-maker decides anything beyond it.
- 5–15 min
Clarify the problem and intended impact
The Feature Partner adds real examples. The Product Owner confirms the value goal.
- 15–30 min
Agree on behavior
Record rules, examples, counterexamples, and boundaries in the versioned source spec.
- 30–40 min
Review implementation and measurement
The team and Measurement Partner agree on tests, risks, and events for impact measurement.
- 40–45 min
Prepare the work
Assign an owner and deadline to critical questions; arrange a focused Spec Clinic when needed.
Concrete outcome
Next verifiable change, investment boundary, acceptance criteria, and open clarification tasks.
Shopping cart example: The shopping cart supports quantity changes and clear shipping costs. Discount codes and a wishlist are out of scope.
Open the related promptWhen clarification is needed · 30 minSpec ClinicBuild · Value
Make a conflicting or oversized spec clear enough to decide on and implement.
Product Owner, relevant developers, and Feature Partner; Measurement Partner when measurement is affected. The Loop Guide facilitates when needed.
AI supportScout flags gaps. Planner highlights effort drivers. Reviewer provides counterexamples. People decide.
The specific conflict or budget risk, affected spec, effort spent, and remaining budget.
Connection to the loopsB1 · B3 · B4 · B6 · B9 · V2
- 0–5 min
Make the bottleneck visible
Show the findings and remaining budget. Which behavior or evidence is still unclear?
- 5–12 min
Identify the essential value
The Product Owner and Feature Partner define the smallest useful, safe, and coherent core.
- 12–22 min
Shape the scope
Defer edge cases and convenience features, simplify variants, or reshape the steps. Quality, safety, and mandatory requirements remain in place.
- 22–30 min
Confirm the spec and next action
Version the updated rules, tests, and measurement plan. Within the mandate, the Product Owner decides whether to continue with a smaller scope or stop. Changes to binding goals or the investment boundary go to the appropriate decision-maker with a recommendation.
Concrete outcome
A confirmed smaller spec or stop decision, updated tests, a clear next action, and an owner.
Shopping cart example: A requested wishlist would exceed the shopping cart budget. Defer it while keeping price accuracy, stock limits, and clear feedback as requirements.
Open the related promptWhen ready for acceptance · reserve 20 minAcceptance SlotBuild
Confirm a completed small change promptly and describe any acceptance gaps clearly.
Feature Partner or named delegate and relevant developer; Product Owner for new requirements.
AI supportThe Reviewer prepares acceptance-criteria IDs, test evidence, a test version, data, and any open manual checks.
An accessible test version, suitable roles and test data, completed criteria, and known limitations. For critical components, also provide evidence of agreed observability, useful alerts, and tested failure behavior. Scale this to criticality, potential harm, frequency of use, dependencies, and recoverability.
Connection to the loopsB8 · B9 · B10
- Beforehand
Prepare the review package
The team provides test access, scenarios, and evidence, including monitoring, alerts, and failure behavior for critical components. Keep evidence lightweight for low-risk changes. The Feature Partner confirms availability.
- 0–5 min
Confirm the scope
Check together which spec version and criteria are being accepted. For critical components, monitoring and useful alerts are part of implementation quality. The agreed operational evidence forms part of the completion criteria.
- 5–15 min
Test real scenarios
The Feature Partner tests important user flows and counterexamples; the developer helps with questions.
- 15–20 min
Record the decision
Record acceptance or a specific gap with an owner. Prioritize new requests separately and take unclear rules to the Spec Clinic. Small, low-risk tasks can be accepted asynchronously; urgent fixes do not wait for this ritual.
Concrete outcome
Business acceptance or a precise correction task. Production approval follows the agreed rules.
Shopping cart example: The Feature Partner changes a quantity, removes the final item, and checks the transition to free shipping.
Open the related promptEvery four weeks · 30 minValue ReviewValue
Every four weeks, evaluate impact, open measurement questions, and relevant operational trends. Agree on the next evidence needed even when data is missing. Address urgent findings immediately.
Product Owner, Delivery Team, and relevant Feature and Measurement Partners.
AI supportThe Impact Analyst summarizes verified Scorecards, user feedback, and open measurement questions.
Comparable data, data-quality checks, guardrail metrics, user feedback, and known uncertainty.
Connection to the loopsV4 · V5 · V6
- 0–5 min
Confirm the data foundation
The Measurement Partner explains which cases are ready for a decision and where evidence is missing.
- 5–20 min
Interpret the impact
The team and Feature Partner consider improvements, side effects, and alternative explanations.
- 20–30 min
Make the product decision
Within the agreed mandate, the Product Owner decides whether to keep, expand, change, or investigate further. Findings requiring a strategic decision go to the responsible sponsors with a recommendation. Record the owner and review date.
Concrete outcome
A reasoned product decision or specific investigation plan. Changes feed back into the selection of the next work.
Shopping cart example: More checkout starts are a signal. The Measurement Partner checks group assignment and error rates before the Product Owner decides whether to expand.
Open the related promptEvery two weeks · 45 min as a starting pointRetrospectiveLearning
Improve collaboration and ways of working; evaluate the ongoing experiment.
Delivery Team and Loop Guide; Product Owner as relevant to the topic. Include personal perspectives.
AI supportThe Learning Guide summarizes only approved observations and the current experiment status.
Team feedback, rework, waiting time, and evidence about AI skills, AI context, and the previous experiment.
Connection to the loopsL1 · L2 · L3 · L5
- 0–5 min
Check in and look back
Make the previous experiment and its open questions visible.
- 5–15 min
Add observations
The team corrects the data and adds its experience of workload and collaboration.
- 15–30 min
Investigate one cause
Choose a pattern and consider evidence and alternative explanations.
- 30–40 min
Design one experiment
The team agrees on the change, owner, expected value, and review date. It might involve an AI skill or a smaller AI context.
- 40–45 min
Confirm the agreement
People confirm the decision. Record confidentiality boundaries and the next evidence needed.
Concrete outcome
An evaluated experiment and at most one new active experiment. No individual performance rankings.
Shopping cart example: As a trial, the Reviewer receives an AI skill for cent-based calculations and stock limits instead of a long, unstructured prompt package.
Open the related promptMonthly or after a proven finding · 30 minLearning ExchangeLearning · Build · Value
Share verified learning so other teams can trial it deliberately.
Representatives of interested teams, AI skill owners, and Loop Guides; Feature or Measurement Partners when needed.
AI supportThe Learning Guide prepares evidenced before-and-after examples. The Reviewer supports reusable test cases.
An evaluated experiment, comparison cases, limits of transferability, and a versioned AI skill or proposed AI context.
Connection to the loopsL5 · L6 · B4 · V2
- 0–5 min
Show the problem and starting point
The contributing team explains the original case and previous way of working.
- 5–15 min
Demonstrate the change with evidence
Compare AI skill versions, context selection, tests, rework, and effort. Be clear about uncertainty.
- 15–25 min
Check transferability
Other teams assess differences and choose a bounded local trial if interested.
- 25–30 min
Agree on accountability
Set an owner, version, comparison cases, and review date for the local trial.
Concrete outcome
A useful learning package with evidence and limitations; a voluntary, verifiable trial in another team.
Shopping cart example: One team shares its verified shopping cart AI testing skill. Another adapts the pricing and stock rules to its working context and tests it again.
Open the related prompt