The LLM's Job Isn't the Answer — It's the Translation

Precise, independently verifiable benchmark figures, an accurately described LLM-to-structured-query mechanism, and a genuinely useful generalizable insight about LLM-as-translator vs. LLM-as-executor — this is substantive, well-researched content with a real conceptual payoff.

The framing device ("for decades, search has worked one way") and the closing "pattern I think we'll see everywhere in applied AI" are stylistic scaffolding — reasonable interpretive commentary, but not independently verified claims, just the author's generalization from one example.