How to Build a Research Agent Army

Nikhil Krishnan's takeaway from healthcare LLM workshops is simple: the first skill is not a workflow, it's learning to ask Claude for help when you're stuck. That means taking a screenshot, asking what to do next, using the model to improve the process itself, and pushing for explanations that match how you think. It is the same muscle as learning to Google well. The real lesson for readers is that AI becomes useful faster when you treat it like a debugging partner, not a person you wait on for answers.