Judging from the project description, Strot may adopt a two-stage architecture of intent extraction and logic solidification: first an LLM analyzes input-output pairs to induce decision rules or a small classifier, and then the rules are compiled into locally executable code.