Workflow Tips

Practical advice for integrating AI tools into your daily development workflow.

Start with Clear Intent

Before prompting an AI tool, know what you want:

• What problem are you solving?
• What does "done" look like?
• What constraints exist?

Use Version Control

Always commit before starting an AI-assisted task. This gives you a clean rollback point if the generated code doesn't work out.

• Git Cheat Sheet
• Make use of Git worktree

Incremental Changes

Make small, testable changes rather than generating large amounts of code at once:

1. Generate a small piece of functionality
2. Review and test it
3. Commit if it works
4. Move to the next piece

Know When NOT to Use AI

AI coding tools are not always the right choice:

• Security-critical code — Crypto, auth logic, and access control deserve extra human attention
• Performance-critical paths — Profile first, then optimize deliberately
• Domain-specific logic — The AI may not understand your business rules
• When you need to learn — Using AI to skip understanding leads to fragile codebases

For risk controls, see Safety and Risk Management.

Pair with AI, Don't Defer to It

Think of AI as a junior developer with broad knowledge but no judgment. You provide the direction, review the output, and make the final call.

Sources: Git tutorial and workflow basics, Trunk-based development (small, incremental changes)