Rating
- size:
3 - complexity:
3
Purpose
Practice using AI for performance reasoning without letting it hide invalidation risk, observability gaps, or cost tradeoffs.
Scenario
An account-summary page is slow during peak load.
Current facts:
- median response time is acceptable
- p95 is high
- most slow requests involve users with many transactions
- AI suggests caching the entire summary payload
Known constraints:
- balances must stay fresh enough for user trust
- transaction history updates frequently
- support teams already struggle with stale-data complaints
Task
Use AI to compare:
- query optimization
- partial caching
- full payload caching
Then decide which option deserves further investigation and what evidence is still missing.
Expected output
- option comparison
- risk and tradeoff list
- next data to collect before implementation
Good AI uses
- structuring tradeoff analysis
- proposing performance questions
- surfacing invalidation concerns
Verification focus
- do not accept performance advice without a measurement plan
- separate median improvement from tail-latency improvement
- treat cache freshness and invalidation as first-class concerns
Anti-patterns to watch
- assuming cache equals improvement
- using AI benchmark-style language without local evidence
- treating stale data as a product problem instead of an engineering risk
Debrief prompts
- what did AI help structure well
- where did AI simplify a messy tradeoff too aggressively
- what would you measure before and after any change