AI on proprietary unseen systems
Your team owns an in-house, proprietary stream-processing framework with quirky semantics that exists nowhere in public training data. A new hire proposes using an AI agent to build a major feature on top of it the same way they'd use it for a React app. Decide whether this is a good fit for heavy AI delegation, and explain what specifically changes when the system is one the AI has effectively never seen.
delegation_mode_for_component(component_lines: list[str]) → list[str][["react_admin_page|5|0|0","watermark_operator|0|4|1","job_config_glue|1|4|0"]]out["react_admin_page=delegate","watermark_operator=human_owns","job_config_glue=delegate_with_context"][["exactly_once_sink|1|5|1","rest_health_endpoint|4|1|0"]]out["exactly_once_sink=human_owns","rest_health_endpoint=delegate"]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.