Umbraco Platinum Partner
We used an AI coding assistant while updating CountThings for Umbraco 17. The useful lesson was not “AI wrote the package”. It was how much friction disappeared when the assistant could work in the codebase and help investigate version changes.
The challenge
The package already supported multiple Umbraco versions, but the build and test pipeline had not been touched for a while. Moving to a new LTS meant checking breaking changes without introducing regressions.
Where the assistant helped
- Getting back up to speed on an older codebase
- Researching version differences while staying in the development workflow
- Spotting small implementation details that were easy to miss
- Automating repetitive test and reinstall steps
What still needed a developer
The assistant accelerated investigation and implementation, but validation still depended on understanding the package, testing the behaviour and deciding whether the changes were correct.
The practical takeaway
AI coding tools are most useful when they reduce context switching and repetitive work without replacing engineering judgement.
