AI allocation by task
You're planning a quarter for a 6-person team. The roadmap has five chunks: (1) a green-field internal analytics dashboard, (2) a tricky migration of a 12-year-old billing state machine to a new schema, (3) bulk-generating 200 CRUD endpoints from an existing OpenAPI spec, (4) a novel rate-limiting algorithm with no clear reference implementation, and (5) routine dependency upgrades across 30 services. Decide for each whether you'd lean heavily on AI agents, use AI as an assistant under close human control, or keep it largely human-led — and explain the principle that drives the split, not just the verdicts.
choose_delegation_mode(spec_clarity: list[int], verification_cost: list[int], no_reference: list[bool]) → list[str][[3,1,5,2,4],[2,5,1,4,1],[false,false,false,true,false]]out["ai_assisted","human_led","agent_heavy","human_led","agent_heavy"][[5],[1],[true]]out["human_led"][[5],[3],[false]]out["ai_assisted"]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.