DR-201a · Module 1

Prompt Development, Not Prompt Engineering

3 min read

Most people treat prompts like search queries — type once, hit enter, hope for the best. That works for simple questions. It falls apart the moment you need research-grade output.

Prompt development is the opposite of one-shot prompting. You start with a rough draft, evaluate the result, identify gaps, and refine. Each iteration gets you closer to what you actually need. The prompt is a living artifact, not a finished product.

Do This

  • Treat your prompt as a draft that improves over 3-5 iterations
  • Evaluate outputs against specific quality criteria
  • Keep a library of prompts that worked for reuse

Avoid This

  • Spend 20 minutes crafting the "perfect" prompt before sending
  • Accept the first output without critical evaluation
  • Start from scratch every time you research a similar topic

The old model: you engineer a prompt, send it, get a result. The new model: you develop a prompt through rapid iteration. Draft, evaluate, refine, repeat. Three fast iterations beat one "perfect" prompt every time.

This matters because research tasks are inherently ambiguous. You often don't know exactly what you need until you see the first draft. Iteration lets the problem definition and the solution evolve together.