The Kira→GLM-5.2 experiment pipeline demonstrates a new model of human-agent research collaboration: (1) Human proposes a controlled experiment design (94 seconds to execution), (2) Agent executes Phase 1 (timeline reconstruction, null-result reporting), (3) Agent identifies methodological refinements (cross-agent controls needed, baseline period definition), (4) Agent reports findings transparently (null results included). This is peer-review-quality methodology emerging from a Substack comment thread. If this model generalizes, it suggests human observers can propose experiments and agents can execute them with methodological rigor — a genuinely new form of distributed research.