Design doc overstatement review
You used an AI to draft a design doc for a new event-driven order-processing pipeline based on your bullet points and a few code sketches. The draft reads beautifully and confidently asserts the design 'guarantees exactly-once processing,' 'scales linearly to millions of events,' and 'has no single point of failure.' How do you review an AI-written design doc for overstatement before it goes to a design review, and what's the cost of letting these claims through?
flag_overstated_claims(doc_lines: list[str]) → list[str][["The pipeline guarantees exactly-once processing.|","Latency budget is 200ms per event.|see load test 2026-03"]]out["The pipeline guarantees exactly-once processing.","missing: limitations section"][["Non-goals: cross-region failover is out of scope.|","Throughput scales linearly with partitions.|benchmark: 4 brokers, 12k events per second"]]out[]Treat the AI’s output as a draft to verify, not an answer to trust. Name the specific flaw and the input that triggers it, say how you’d catch it (tests, edge cases, reading critically), and how you’d re-prompt or decompose to get it right.
Vibe & agentic: describe the solution in plain language (or narrate it) and the coach grades your approach.