Step 3 - Generate a Plan

With a refined requirement and a technical spec in place, we move to planning.

The goal of this step is to produce a small, executable plan that an AI-native IDE can help you implement safely.

Inputs

To generate a useful plan, you should have:

• The refined requirement from Step 1.
• The technical spec from Step 2.
• A list of relevant files and modules that will likely be touched.

Using AI to Draft the Plan

In your AI-native IDE or planning tool:

1. Provide the requirement and spec.
2. Attach or reference key files.
3. Ask the AI to "Create a step-by-step plan to implement this spec. Break it into small, ordered steps tied to specific files or modules. Include necessary tests."

The output should look like a checklist, not a wall of prose.

Shaping the Plan

Do not accept the first draft blindly. Review and refine:

• Is the order sensible (e.g., backend before frontend, contracts before consumers)?
• Are steps small enough to implement and review in isolation?
• Is anything missing (migrations, feature flags, docs, monitoring)?

If needed, ask the AI to:

• Split large steps into smaller ones.
• Group related changes into phases.
• Flag risky steps that might require extra testing or review.

Representing the Plan

For complex work, keep the plan in a simple, versioned artifact (for example, a markdown checklist linked from the ticket):

• [ ] Add new column to invoices table and migration.
• [ ] Update domain model and validation logic.
• [ ] Extend API endpoint and types.
• [ ] Add unit tests for new rules.

This makes it easy to:

• Track progress.
• Re-sync with stakeholders.
• Hand off work between engineers if needed.

The plan is living. You will adjust it as you learn more in Steps 4–5, but you should always know what the next 1–3 steps are.