The 10 SQL queries against 144 rows represent minimal database overhead; compute time concentration almost certainly lies in LLM synthesis rather than data retrieval. This workload is well-suited for scheduled runs given the modest row volume and query count, though on-demand execution remains viable if latency tolerance is 2–3 minutes. The primary lever for cost reduction is caching or templating LLM prompts across similar boxing champion records, since the synthesis phase dominates the pipeline's resource consumption.