Investigate build-time regression in 2.14 asset pipeline
Dedicate one perf engineer for two weeks to bisect and land a fix; ship macOS 15 runner image update in parallel.E-06E-10
Five stages, in the order they ran: what Juno planned to find out, what it retrieved, what it concluded, what it checked before showing you anything, and only then the recommendation itself.
- Identify customer evidence
No customer interview on file for this theme.
- Review support trends
1 support tickets grouped by theme.
- Review sales feedback
No sales note references this theme.
- Review engineering feasibility
Engineering signal retrieved and reflected in the technical risk factor.
- Compare against strategic goals
Protects the reliability promise rather than advancing a growth goal.
- Name what is still unknown
1 gaps recorded with an owner and a discovery question.
- Check for conflicting stakeholder input
No stakeholder disagreement found in the retrieved artifacts.
- E-061 artifact
Median build times regressed 38% since 2.14; correlated with asset pipeline.
- E-101 artifact
macOS 15 build failures affect 6 customers this week.
Retention risk is real but no account has attached a dollar figure.
Sustained slowdown affects every paid workspace that builds daily.
Protects the reliability promise rather than advancing a growth goal.
- The regression may be runner image drift rather than the asset pipeline, which would invalidate the two-week estimate.
- macOS 15 linker failures could be an unrelated second issue absorbing the same engineer.
- Asset pipeline is the cause, not runner image drift.
Ranks third on strong engineering evidence and clear customer pain, but the revenue exposure is unquantified.
Build times regressed 38% since 2.14, correlated with the new asset pipeline. Compounded by intermittent macOS 15 linker failures affecting 6 customers.
High evidence strength but smaller revenue lever than SSO or onboarding.
Recommendation quality
How well supported this recommendation is, measured against the evidence Juno could actually retrieve.
2 of 6 enterprise source types contributed at least one artifact.
Engineering Alert, Support Ticket
Retrieved stakeholders agree on the problem statement.
Whether every section of the brief could be filled from retrieved evidence.
No recommendation advances to the roadmap without an explicit decision.
Findings resting on one artifact are held at lower strength rather than stated with certainty.
Supporting evidence
Evidence strength reflects how many independent artifacts support this recommendation and how varied they are, not how certain the model sounds.
| Finding | Artifact | Severity | Corroboration |
|---|---|---|---|
E-06Fact Median build times regressed 38% since 2.14; correlated with asset pipeline. | Engineering Alert Eng Slack: build-time regression | High severity | High confidence 1 artifact |
| Finding | Artifact | Severity | Corroboration |
|---|---|---|---|
E-10Fact macOS 15 build failures affect 6 customers this week. | Support Ticket Ticket #48622: build failures on macOS 15 | High severity | High confidence 1 artifact |
Transparent scoring model
Risks
- The regression may be runner image drift rather than the asset pipeline, which would invalidate the two-week estimate.
- macOS 15 linker failures could be an unrelated second issue absorbing the same engineer.
Additional discovery recommended
Juno names what it does not know, who can answer it, and how to ask, so limited confidence turns into a discovery step rather than a dead end.
Why it matters: A concentrated regression is a targeted fix; a uniform one is a pipeline redesign.
- Is the regression uniform across runner sizes or concentrated on specific configs?
Recommended next step
Bisect 2.14 asset pipeline against 2.13 baseline
Hypothesis: Regression isolates to the asset pipeline change, not runner drift.
Juno proposes this step because it would close the largest gap listed above. You decide whether to run it.
Decision required. Juno recommends, you decide, and nothing reaches the roadmap without you.
Final prioritization decisions remain the responsibility of the Product Manager.