Start with the molecular research problem.

The most useful first conversation explains the decision to improve, the current workflow, the data boundary, and what evidence would make an evaluation worthwhile.