Summary
This episode is an unscripted demonstration of using the Anthropic Claude AI agent to refactor a C++ codebase from using exceptions to std::expected. Jason walks through the entire process showing how the AI analyzes the code, creates error types, modifies function signatures, handles error propagation, and maintains test coverage. While requiring some guidance and iteration, the AI successfully completed the complex refactoring in about an hour for under $10, demonstrating that modern AI tools have moved far beyond simple autocomplete for C++ development.
Related C++ Standard Sections
This episode covers topics found in these sections of the C++ standard:
-
[expected]40% match