I am moving further away from just telling an AI to 'fix' something, instead ive started using it for docs and codebase analysis that runs while I code myself.

Agentic coding helps us move faster — PR reviews are quick, I still check edge cases manually, but raising a PR and deploying usually happens the same day.

Manual testing is more important than ever.