- Traditional software development involves a stable branch and an unstable branch, with a freeze and testing period before release.
- AI coding changes both development and usage, enabling users to modify software with AI agents and making code more malleable.
- Repositories can serve as templates for users to specialize, rather than just finished products, allowing adaptation to specific needs.
- Projects like DwarfStar benefit from providing solid examples that coding agents can extrapolate to new configurations and models.
- Documentation should be written for both humans and coding agents to facilitate understanding and modification of the system.
- The release process may become more fluid, with many experimental branches as integral parts of the project.