Interview of Julien Dubois
summary
GitHub Copilot handles boilerplate and quick method generation because it’s trained on public GitHub data, while JHipster produces fully tested, reproducible scaffolding with hundreds of configuration files and continuous checks, so use JHipster for deterministic project bootstraps and GitHub Copilot to fill specific business methods; LLM limits include token context windows, drifted training data (often ~two years old) and weaker refactoring, so add retrieval/RAG or prioritized docs; new Dependabot + Gradle integration exports full dependency trees and flags transitive updates.
AI-written summary · about
chapters (11)
- 0:00Welcome to Julien Dubois
- 0:53GitHub Copilot and AI pair programming
- 2:01JHipster and AI tools
- 3:47Can AI replace JHipster?
- 5:52Custom AI models for frameworks
- 7:15Bootstrapping vs AI generation
- 9:00Open source in AI training
- 10:48AI model limitations
- 12:14Updating dependencies with Dependabot
- 13:40Gradle and Dependabot integration
- 14:22Closing remarks
related talks
53mCopilot in Your Java Tooling: From CLI to SDK to Plugins
01h 20mGithub Copilot - The World's Most Widely Adopted AI Developer Tool
40mUnlocking Productivity with GitHub Copilot Best Practices and Tips
24mGitHub Copilot Agents + Sonar | The Blueprint for Agentic Development
50mGitHub Copilot: Coding Will Never Be The Same Again
48mCode Generation in 2024: How AI Will Take Your Job…to the Next Level!
31mGithub Co-Pilot: Addictive or Effective?
49mGithub Copilot vs. Amazon CodeWhisperer For Java Developers