Work through what is failing and find a useful next experiment.
Rigorous = assumptions, incentives, and failure modes.
Tell me what's failing and what you've already tried.
SmallAI