Step 1 - Define & Clarify

The biggest enemy of velocity is ambiguity. If you give an AI a vague requirement, it will generate vague code that looks plausible but fails to solve the real problem.

Step 1 is about pre-code validation: making sure we are solving the right problem before we touch implementation.

The Process

1. Start from the source artifact
   Take the ticket, spec, or request from your tracker or document. Read it end to end.

2. Use an AI "clarity assistant"
   In your AI-native IDE or chat tool, start a focused session whose only job is to interrogate the requirement. For example:

   • Paste the ticket or problem statement.
   • Ask: "Analyze this task for logical gaps, edge cases, and missing requirements. Ask me questions to clarify."

3. Iterate on questions and answers
   Treat the AI as a collaborator that helps you think, not as a source of truth.

   • Answer questions.
   • Push back where it overcomplicates things.
   • Ask it to propose alternative framings of the same requirement.

4. Produce a Refined Requirement Statement
   When the conversation stabilizes, ask the AI to draft a concise, unambiguous description of the task. Edit it yourself until you are confident it captures:

   • The goal.
   • The main flows.
   • Notable edge cases.
   • Out-of-scope items.

The output of Step 1 is a refined requirement you can attach to the ticket and reuse in later steps.

Why This Matters

Across real projects, most defects originate from bad or incomplete requirements, not from syntax errors. Catching gaps here:

• Reduces rework later.
• Makes Step 2 (Technical Spec) much faster.
• Gives product and delivery something concrete to react to and improve.

Step 1 is cheap compared to rewriting features after they ship. Do not skip it.