In his video series ‘Levels of Enlightenment,’ Rob Bos shares his journey and practical lessons learned while using GitHub Copilot over the last year and a half. Each episode highlights an “aha” moment, outlining strategies to help developers extract greater value from Copilot and improve their software development process.

Overview

Rob Bos’s series documents his evolving approach to Copilot, breaking down the following key lessons:

1. Explain What You Want to Achieve

By getting more descriptive with code and clearly stating intentions, users can enable Copilot to generate more relevant and accurate suggestions. This video demonstrates how adjusting the way code is written prompts better results.

2. Know Your Context

Understanding how Copilot leverages code context allows users to tailor their environment and cues, resulting in improved code completions. Rob shows how deliberate context management can enhance Copilot’s effectiveness.

3. Copy Method Calls as Comments

Rob illustrates how copying method calls into comments before implementing them can accelerate code creation, providing Copilot with explicit cues for the intended functionality.

4. Top-Down Programming Instead of Bottom-Up

Shifting from bottom-up to top-down programming helps developers work more efficiently with Copilot. Planning code at a higher level guides the AI and leads to more cohesive suggestions.

5. Typos in Your Prompt Do Not Matter

Rob encourages letting go of perfectionism regarding prompt typos; Copilot often interprets them correctly, allowing for faster coding without worrying about minor errors.

6. Use the Chat

Leveraging Copilot’s chat interface can provide faster results. Rob explains how frequent use of the chat function allows developers to focus on their goals rather than syntax details.

7. #file is Your Biggest Friend

Being explicit by using directives like #file and #selection in prompts provides Copilot with precise context, which improves the accuracy and usefulness of its outputs.

8. Accept the Truth

Rob advises embracing the nature of Copilot as an AI based on Large Language Models: it can be non-deterministic and not always correct. Developers remain in control and should guide and validate Copilot’s outcomes.

9. Be Smart

Finally, Rob discusses the importance of creativity when using Copilot. Developers should explore new use cases, delegate more work, and allow Copilot to aid in exploring novel coding approaches.

Conclusion

Rob Bos’s ‘Levels of Enlightenment’ offers a comprehensive roadmap for maximizing GitHub Copilot’s capabilities. Each lesson provides actionable insights for developers looking to integrate AI tools more thoughtfully and productively into their workflows.

This post appeared first on Rob Bos’ Blog. Read the entire article here