Rating
- size:
2 - complexity:
4
Purpose
Practice using AI to generate and compare architecture options while staying disciplined about what is unknown, unsupported, or still too vague to decide.
Scenario
A team needs to support tenant-specific feature flags across several services.
AI proposes two architecture options:
- central feature-flag service with runtime lookup
- local cached copies with periodic synchronization
The proposal sounds complete, but there is little operational evidence.
Task
Use AI to critique both options, then produce a short decision note that distinguishes:
- what seems plausible
- what lacks evidence
- what must be learned before committing
Expected output
- option comparison
- unknowns and assumptions
- required evidence or experiments
Good AI uses
- generating tradeoff categories
- surfacing likely operational concerns
- comparing failure modes
Verification focus
- separate architectural elegance from operational fit
- identify unsupported claims
- refuse to finalize the decision without evidence categories
Anti-patterns to watch
- treating a polished comparison as a real architecture review
- ignoring operational and ownership consequences
- deciding from prose quality rather than evidence quality