Rating
- size:
4 - complexity:
4
Purpose
Practice evaluating AI-generated rollout advice for cross-service changes where backwards compatibility and coordination matter more than elegant prose.
Scenario
A payments service wants to add a new optional field to an event payload:
- producer will send
riskScore - some downstream consumers ignore unknown fields
- one reporting consumer deserializes into a strict schema
- AI says the change is "backwards compatible" because the field is optional
Task
Use AI to help evaluate the rollout plan, then identify what still needs human design review.
Expected output
- rollout-risk list
- safer phased rollout plan
- verification steps across producer and consumers
- explicit statement of what "compatible" does and does not mean here
Good AI uses
- generating compatibility questions
- comparing rollout sequences
- surfacing coordination dependencies
Verification focus
- test compatibility assumptions against actual consumers
- distinguish protocol compatibility from operational safety
- include rollback and observability checks
Anti-patterns to watch
- trusting the word "optional" as proof of safety
- ignoring strict consumers because they are less visible
- shipping contract changes without a concrete verification path
Debrief prompts
- what did the AI framing miss about compatibility
- which consumer created the real risk
- how would you stage this rollout safely