Learn AI-Assisted Development
Build useful software with AI while keeping your judgment, product thinking, and engineering skills in the driver’s seat.
Learn a repeatable workflow for giving AI the right context, breaking work into safe steps, reviewing generated code, and shipping with confidence. The tools will change; these habits keep working.
Prompt starter kit
Prompts you can use today
Strong prompts make the work easier to review. Replace the bracketed details with your project context, then ask the AI to work in small, verifiable steps.
Plan before you build
Use this when starting a feature or unfamiliar task.
You are a senior software engineer working in an existing codebase. Goal: [describe the feature or outcome] Relevant context: [stack, users, constraints, and links to related files] Before changing code: 1. Identify the files and existing patterns that matter. 2. Explain your understanding and list assumptions. 3. Propose the smallest implementation plan. 4. Call out risks, edge cases, and how we will verify it. Do not edit files until I approve the plan.
Build one vertical slice
Keep implementation focused, reviewable, and easy to test.
Implement only this slice of the feature: [specific user-visible behavior] Follow the existing project patterns and use the current dependencies. Requirements: - Include loading, empty, success, and error states. - Preserve existing behavior outside this slice. - Keep types and validation explicit. - Do not add a dependency unless you explain why it is necessary. After making the change, summarize the files changed and the checks I should run.
Debug with evidence
Turn a vague error into a focused investigation.
Help me investigate this issue without guessing. Expected behavior: [what should happen] Actual behavior: [what happens instead] Error or logs: [paste the exact error] Relevant code: [paste the smallest useful snippet] First list the most likely causes and what evidence would confirm each one. Then recommend the smallest fix, explain why it works, and provide a verification step. Do not rewrite unrelated code.
A modern learning path for AI-assisted coding
Start with the learning path that matches your experience, then use this workflow to turn each lesson into practice:
Understand the problem
Describe the user outcome, constraints, existing behavior, and what must not change.
Ask for a plan
Have the AI inspect the relevant context and propose a small implementation before it edits code.
Build in slices
Implement one user-visible piece at a time, keeping each change easy to review and undo.
Verify and improve
Run checks, test edge cases, review security and accessibility, and refine based on evidence.
Keep your judgment in the loop
AI can draft code quickly, but you remain responsible for the result. Check generated changes against the product goal, tests, privacy requirements, accessibility expectations, and your project’s conventions before shipping.